ALR_Client_Side.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import scipy\n",
    "import scipy.io as sio\n",
    "from scipy import stats\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy\n",
    "import pandas\n",
    "import sklearn\n",
    "from sklearn import preprocessing\n",
    "from sklearn import linear_model\n",
    "from sklearn import metrics\n",
    "from sklearn.cluster import KMeans\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "import tensorflow\n",
    "\n",
    "import math\n",
    "import random\n",
    "import os\n",
    "import time\n",
    "import csv\n",
    "import functools\n",
    "\n",
    "import ee\n",
    "import folium"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pandas_quality_codes(inputs):\n",
    "    inputs = pandas.DataFrame(inputs)\n",
    "    features = inputs.columns.to_list()\n",
    "    \n",
    "    # Generate a list of series containing which decile each of the samples in the series for each feature belong to\n",
    "    feature_partitions = [pandas.qcut(x=values, q=10) for feature, values in inputs.items()]\n",
    "\n",
    "    # Create a dataframe containing the codes for the decile that each feature sample belongs to (from 1 to 10)\n",
    "    feature_deciles = pandas.concat([feature.cat.codes for feature in feature_partitions], axis=1)\n",
    "    feature_deciles.columns = features\n",
    "\n",
    "    # To see which combinations of feature inputs we have we can generate a list of codes where each digit in the code is the decile that each feature sample belongs to from the calibration dataset\n",
    "    quality_codes = list(map(int, (feature_deciles).astype(str).apply(func=tuple, axis='columns').str.join(sep='').unique().tolist()))\n",
    "\n",
    "    # Generate a dictionary mapping each feature name to it's decile ranges contained in the calibration data for later use in classifying pixels in an input image\n",
    "    feature_decile_ranges = dict(zip(features, [dict(zip(range(10), [list(range) for range in feature.cat.categories.to_tuples().values.tolist()])) for feature in feature_partitions]))\n",
    "    \n",
    "    return quality_codes, feature_decile_ranges"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Use the keras backend to define a custom activation function used in the SL2P neural network model\n",
    "def tansig(x):\n",
    "    return (2 / (1 + tensorflow.keras.backend.exp(-2*x)) ) - 1\n",
    "# Add this function as an activation function to this scripts custom keras objects for easier use in models\n",
    "tensorflow.keras.utils.get_custom_objects()['tansig'] = tensorflow.keras.layers.Activation(tansig)\n",
    "\n",
    "# Create a callback for early stopping the model (prevent the model from overfitting) by stopping the training when the mean squared error on the validation dataset does not improve over 25 epochs\n",
    "stop_on_val_loss = tensorflow.keras.callbacks.EarlyStopping(monitor=\"val_loss\", min_delta=0.0001, patience=25, verbose=0, mode=\"min\", restore_best_weights=True)\n",
    "\n",
    "stop_on_loss = tensorflow.keras.callbacks.EarlyStopping(monitor=\"loss\", min_delta=0.0001, patience=25, verbose=0, mode=\"min\", restore_best_weights=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### To-do\n",
    " - Find a Kmeans clustering algorithm that uses the Manhattan distance metric"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 500,
   "metadata": {},
   "outputs": [],
   "source": [
    "# The main goal of clustering the input data is so that during validation we do not just randomly sample validation data, \n",
    "# or just sample a subset of the data in a specific range of the output values. Instead we can now force the model to extrapolate during validation by removing entire chunks of data \n",
    "# (removing 20% of the clusters of data as opposed to 20% of the data randomly).\n",
    "def cluster_training(inputs, outputs, cluster_labels: numpy.ndarray, keras_model, batch_size: int = 32, epochs_per_loop: int = 100, verbosity: int = 1, callbacks: list = [], num_batches: int = 100, validation_split: float = 0.2, monte_carlo_loops: int = 1):\n",
    "    inputs = pandas.DataFrame(inputs)\n",
    "    features = inputs.columns\n",
    "    outputs = pandas.Series(outputs)\n",
    "    output_var = outputs.name\n",
    "    \n",
    "    num_samples = inputs.shape[0]\n",
    "    num_features = inputs.shape[1]\n",
    "    \n",
    "    if (len(cluster_labels) == num_samples):\n",
    "        cluster_min = cluster_labels.min()\n",
    "        cluster_max = cluster_labels.max()\n",
    "        num_clusters = len(numpy.unique(cluster_labels))\n",
    "        \n",
    "        sample_clusters = cluster_labels\n",
    "    else:\n",
    "        # Sets the number of clusters to a minimum of 100, or higher if the user requests more based on the number of samples\n",
    "        cluster_min = 0\n",
    "        cluster_max = max(100, num_samples/(num_batches*1000))\n",
    "        num_clusters = cluster_max\n",
    "        # Cluster the input data using euclidean distances between data points (not currently possible to specify other distance metrics)\n",
    "        # Getting the labels for each of the points used to fit the KMeans clusterer\n",
    "        sample_clusters = KMeans(n_clusters=num_clusters).fit(inputs).labels_\n",
    "    \n",
    "    # Dataframe that stores all of the inputs and the outputs to cluster the inputs alongside their respective outputs using the cluster labels\n",
    "    input_data = inputs.join(outputs)\n",
    "    # Get a dictionary mapping the cluster number to the sample indices in that cluster\n",
    "    cluster_dictionary = input_data.groupby(sample_clusters).groups\n",
    "    \n",
    "    num_validation_clusters = int(math.ceil(num_clusters*validation_split))\n",
    "    \n",
    "    # The implementation of the loop and the generation of a randomly subsampled set of validation data (and calibration data) in each loop,\n",
    "    # where in each loop the model is subsequently trained on the calibration data and validated on the validation data.\n",
    "    # This is a form of Monte-Carlo cross-validation applied to training our neural network.\n",
    "    for loop in range(monte_carlo_loops):\n",
    "        \n",
    "        validation_clusters = random.sample(range(cluster_min, cluster_max), num_validation_clusters)\n",
    "        validation_samples = [cluster_dictionary[cluster] for cluster in validation_clusters]\n",
    "        validation_data = pandas.concat([input_data.loc[cluster] for cluster in validation_samples], ignore_index=True)\n",
    "        num_validation_samples = len(validation_data)\n",
    "        \n",
    "        validation_set = (preprocessing.scale(validation_data[features]), validation_data[output_var].values)\n",
    "        \n",
    "        \n",
    "        calibration_clusters = list(set(range(cluster_min, cluster_max)) - set(validation_clusters))\n",
    "        calibration_samples = [cluster_dictionary[cluster] for cluster in calibration_clusters]\n",
    "        calibration_data = pandas.concat([input_data.loc[cluster] for cluster in calibration_samples], ignore_index=True)\n",
    "        num_calibration_samples = len(calibration_data)\n",
    "        \n",
    "        training_data = preprocessing.scale(calibration_data[features])\n",
    "        training_targets = calibration_data[output_var].values\n",
    "\n",
    "        \n",
    "        keras_model.fit(x=training_data, y=training_targets, batch_size=batch_size, epochs=epochs_per_loop, verbose=verbosity, callbacks=callbacks, shuffle=True, validation_data=validation_set)\n",
    "        \n",
    "        print(\"Completed monte carlo training loop {} of {}\".format(loop+1, monte_carlo_loops))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_prediction(true_values, predicted_values, ideal_predictor_range: list = [0,1], description: str = \"\"):\n",
    "    # Stacks the arrays (true values and predicted values) into a 2d array\n",
    "    stacked_data = numpy.vstack([true_values, predicted_values])\n",
    "\n",
    "    # Get the density of the data at each sample point\n",
    "    density_data = stats.gaussian_kde(stacked_data)(stacked_data)\n",
    "\n",
    "    # Reorder the data to be plotted so higher density points are plotted on top\n",
    "    density_sort_order = density_data.argsort()\n",
    "    x = true_values[density_sort_order]\n",
    "    y = predicted_values[density_sort_order]\n",
    "    c = density_data[density_sort_order]\n",
    "\n",
    "    # Calculate the RMSE of the model\n",
    "    rmse = metrics.mean_squared_error(x, y, squared=False)\n",
    "\n",
    "    # Using matplotlib to plot the predictions against the true values, with a line showing the ideal behaviour of any model\n",
    "    ideal_predictor = numpy.linspace(ideal_predictor_range[0], ideal_predictor_range[1], 1000)\n",
    "\n",
    "    fig, ax = plt.subplots(1, 1, figsize=(15,15))\n",
    "\n",
    "    ax.scatter(x, y, c=c)\n",
    "    ax.plot(ideal_predictor, ideal_predictor, c='r')\n",
    "    ax.set_title((description+\" - RMSE: {}\").format(rmse))\n",
    "    \n",
    "    plt.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "def cluster_RMSE(inputs, outputs, predictions, band_errors):\n",
    "    features = inputs.columns\n",
    "    output_var = outputs.name\n",
    "    prediction_name = predictions.name\n",
    "    \n",
    "    prediction_dataset = inputs.join(outputs).join(predictions)\n",
    "    \n",
    "    const_frame = pandas.DataFrame(True, index=range(0,inputs.shape[0]), columns=[output_var, prediction_name])\n",
    "\n",
    "    def sample_RMSE(sample):\n",
    "        samples = sample[1].loc[features]\n",
    "\n",
    "        range_min = samples.sub(band_errors)\n",
    "        range_max = samples.add(band_errors)\n",
    "\n",
    "        indistinguishable = prediction_dataset[((inputs >= range_min) & (inputs <= range_max)).join(const_frame)].dropna()\n",
    "\n",
    "        num_indistinguishable = len(indistinguishable)\n",
    "        true_values = indistinguishable[output_var]\n",
    "        prediction_set = indistinguishable[prediction_name]\n",
    "\n",
    "        sample_RMSE = ((prediction_set.sub(true_values).pow(2).sum())/num_indistinguishable) ** 0.5\n",
    "        return sample_RMSE\n",
    "    \n",
    "    RMSE_list = list(map(sample_RMSE, prediction_dataset.iterrows()))\n",
    "    \n",
    "    return numpy.asarray(RMSE_list)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### To-do\n",
    "- Adjust inputs to add noise to data (approximately 5% added noise on the reflectances) - more realistic training of neural net"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 505,
   "metadata": {},
   "outputs": [],
   "source": [
    "# File path to the simulated data to traputs the neural network with, relative to the file location of the script\n",
    "mat_file_path = \"data/s2_sl2p_weiss_or_prosail_inout.mat\"\n",
    "variables_to_load = [\"Input\", \"Output\"]\n",
    "\n",
    "# Load matlab data into dictionary mapping the variable_names to arrays containing their values\n",
    "matlab_data = sio.loadmat(file_name=mat_file_path, variable_names=variables_to_load)\n",
    "\n",
    "# Load the simulated reflectances and the angles into dataframes\n",
    "input_reflectances = pandas.DataFrame(data=matlab_data['Input']['Rho_Toc'][0][0])\n",
    "input_angles = pandas.DataFrame(data=matlab_data['Input']['Angles'][0][0])\n",
    "\n",
    "# Format the dataframe to contain the correct column labels with all of the input data\n",
    "inputs_SL2P = pandas.DataFrame(pandas.concat([input_reflectances, input_angles], axis=1, ignore_index=True))\n",
    "cluster_labels_SL2P = matlab_data[\"Input\"][\"Cats\"][0][0].flatten()\n",
    "\n",
    "bands_SL2P = ['B4', 'B5', 'B6', 'B7', 'B8A', 'B9', 'B11', 'B12']\n",
    "angles_SL2P = ['SZA', 'SAA', 'ALT']\n",
    "features_SL2P = bands_SL2P + angles_SL2P\n",
    "inputs_SL2P.columns = features_SL2P\n",
    "\n",
    "SL2P_quality_codes, SL2P_input_deciles = pandas_quality_codes(inputs_SL2P)\n",
    "\n",
    "# Scale the full set of inputs for prediction\n",
    "scaled_inputs_SL2P = preprocessing.scale(inputs_SL2P)\n",
    "\n",
    "# Extract the outputs from the matlab_data dictionary into pandas series\n",
    "LAI = pandas.Series(data=matlab_data['Output']['LAI'][0][0].flatten(), name=\"LAI\")\n",
    "FAPAR = pandas.Series(data=matlab_data['Output']['FAPAR'][0][0].flatten(), name=\"FAPAR\")\n",
    "FCOVER = pandas.Series(data=matlab_data['Output']['FCOVER'][0][0].flatten(), name=\"FCOVER\")\n",
    "\n",
    "output_var = \"LAI\"\n",
    "outputs = pandas.concat([LAI, FAPAR, FCOVER], axis=1)[output_var]\n",
    "\n",
    "SL2P_response_range = [outputs.min(), outputs.max()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 501,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Using Keras to create a sequential model neural network which only has simple dense layers of the specified number of nodes\n",
    "# The below model closely follows the SL2P neural network architecture as described\n",
    "SL2P_model = tensorflow.keras.models.Sequential([\n",
    "    tensorflow.keras.layers.Dense(11, input_shape = [inputs_SL2P.shape[1]]),\n",
    "    tensorflow.keras.layers.Dense(5, activation='tansig'),\n",
    "    tensorflow.keras.layers.Dense(1)\n",
    "])\n",
    "# Compiling the model to minimize the mean squared error loss function and use the NADAM optimizer\n",
    "SL2P_model.compile(\n",
    "    optimizer=tensorflow.keras.optimizers.Nadam(),\n",
    "    loss='mse')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 508,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 32965 samples, validate on 8439 samples\n",
      "Epoch 1/100\n",
      "32965/32965 [==============================] - 2s 62us/sample - loss: 3.6943 - val_loss: 1.3263\n",
      "Epoch 2/100\n",
      "32965/32965 [==============================] - 1s 45us/sample - loss: 1.0345 - val_loss: 0.8033\n",
      "Epoch 3/100\n",
      "32965/32965 [==============================] - 2s 47us/sample - loss: 0.6416 - val_loss: 0.5135\n",
      "Epoch 4/100\n",
      "32965/32965 [==============================] - 1s 44us/sample - loss: 0.4657 - val_loss: 0.4177\n",
      "Epoch 5/100\n",
      "32965/32965 [==============================] - 1s 44us/sample - loss: 0.4039 - val_loss: 0.4008\n",
      "Epoch 6/100\n",
      "32965/32965 [==============================] - 2s 46us/sample - loss: 0.3750 - val_loss: 0.3674\n",
      "Epoch 7/100\n",
      "32965/32965 [==============================] - 1s 43us/sample - loss: 0.3559 - val_loss: 0.4047\n",
      "Epoch 8/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.3429 - val_loss: 0.3452\n",
      "Epoch 9/100\n",
      "32965/32965 [==============================] - 1s 43us/sample - loss: 0.3314 - val_loss: 0.3549\n",
      "Epoch 10/100\n",
      "32965/32965 [==============================] - 1s 43us/sample - loss: 0.3222 - val_loss: 0.3295\n",
      "Epoch 11/100\n",
      "32965/32965 [==============================] - 1s 44us/sample - loss: 0.3135 - val_loss: 0.3110\n",
      "Epoch 12/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.3046 - val_loss: 0.3338\n",
      "Epoch 13/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.2974 - val_loss: 0.2962\n",
      "Epoch 14/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.2884 - val_loss: 0.2912\n",
      "Epoch 15/100\n",
      "32965/32965 [==============================] - 1s 41us/sample - loss: 0.2810 - val_loss: 0.2916\n",
      "Epoch 16/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.2734 - val_loss: 0.2992\n",
      "Epoch 17/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.2669 - val_loss: 0.3038\n",
      "Epoch 18/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.2596 - val_loss: 0.2883\n",
      "Epoch 19/100\n",
      "32965/32965 [==============================] - 1s 43us/sample - loss: 0.2531 - val_loss: 0.2676\n",
      "Epoch 20/100\n",
      "32965/32965 [==============================] - 1s 43us/sample - loss: 0.2491 - val_loss: 0.2606\n",
      "Epoch 21/100\n",
      "32965/32965 [==============================] - 1s 41us/sample - loss: 0.2458 - val_loss: 0.2813\n",
      "Epoch 22/100\n",
      "32965/32965 [==============================] - 1s 43us/sample - loss: 0.2414 - val_loss: 0.2867\n",
      "Epoch 23/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.2388 - val_loss: 0.2744\n",
      "Epoch 24/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.2369 - val_loss: 0.2856\n",
      "Epoch 25/100\n",
      "32965/32965 [==============================] - 1s 43us/sample - loss: 0.2333 - val_loss: 0.2658\n",
      "Epoch 26/100\n",
      "32965/32965 [==============================] - 1s 41us/sample - loss: 0.2319 - val_loss: 0.2646\n",
      "Epoch 27/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.2301 - val_loss: 0.2902\n",
      "Epoch 28/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.2272 - val_loss: 0.3069\n",
      "Epoch 29/100\n",
      "32965/32965 [==============================] - 1s 43us/sample - loss: 0.2267 - val_loss: 0.2843\n",
      "Epoch 30/100\n",
      "32965/32965 [==============================] - 1s 41us/sample - loss: 0.2247 - val_loss: 0.2799\n",
      "Epoch 31/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.2222 - val_loss: 0.2502\n",
      "Epoch 32/100\n",
      "32965/32965 [==============================] - 1s 41us/sample - loss: 0.2202 - val_loss: 0.2688\n",
      "Epoch 33/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.2200 - val_loss: 0.2847\n",
      "Epoch 34/100\n",
      "32965/32965 [==============================] - 1s 41us/sample - loss: 0.2192 - val_loss: 0.3141\n",
      "Epoch 35/100\n",
      "32965/32965 [==============================] - 1s 43us/sample - loss: 0.2178 - val_loss: 0.2537\n",
      "Epoch 36/100\n",
      "32965/32965 [==============================] - 1s 41us/sample - loss: 0.2166 - val_loss: 0.2606\n",
      "Epoch 37/100\n",
      "32965/32965 [==============================] - 1s 43us/sample - loss: 0.2162 - val_loss: 0.2917\n",
      "Epoch 38/100\n",
      "32965/32965 [==============================] - 1s 42us/sample - loss: 0.2136 - val_loss: 0.2940\n",
      "Epoch 39/100\n",
      "32965/32965 [==============================] - 1s 44us/sample - loss: 0.2130 - val_loss: 0.2540\n",
      "Epoch 40/100\n",
      "32965/32965 [==============================] - 1s 45us/sample - loss: 0.2127 - val_loss: 0.2855\n",
      "Epoch 41/100\n",
      "32965/32965 [==============================] - 1s 44us/sample - loss: 0.2117 - val_loss: 0.3308\n",
      "Epoch 42/100\n",
      "32965/32965 [==============================] - 1s 45us/sample - loss: 0.2101 - val_loss: 0.2634\n",
      "Epoch 43/100\n",
      "32965/32965 [==============================] - 1s 43us/sample - loss: 0.2092 - val_loss: 0.2938\n",
      "Epoch 44/100\n",
      "32965/32965 [==============================] - 1s 44us/sample - loss: 0.2093 - val_loss: 0.2735\n",
      "Epoch 45/100\n",
      "32965/32965 [==============================] - 1s 44us/sample - loss: 0.2080 - val_loss: 0.2982\n",
      "Epoch 46/100\n",
      "32965/32965 [==============================] - 1s 43us/sample - loss: 0.2077 - val_loss: 0.2724\n",
      "Epoch 47/100\n",
      "32965/32965 [==============================] - 1s 44us/sample - loss: 0.2070 - val_loss: 0.2498\n",
      "Epoch 48/100\n",
      "32965/32965 [==============================] - 2s 46us/sample - loss: 0.2062 - val_loss: 0.2464\n",
      "Epoch 49/100\n",
      "32965/32965 [==============================] - 2s 47us/sample - loss: 0.2061 - val_loss: 0.2665\n",
      "Epoch 50/100\n",
      "32965/32965 [==============================] - 2s 46us/sample - loss: 0.2058 - val_loss: 0.2788\n",
      "Epoch 51/100\n",
      "32965/32965 [==============================] - 2s 46us/sample - loss: 0.2050 - val_loss: 0.2772\n",
      "Epoch 52/100\n",
      "32965/32965 [==============================] - 2s 47us/sample - loss: 0.2051 - val_loss: 0.2937\n",
      "Epoch 53/100\n",
      "32965/32965 [==============================] - 2s 46us/sample - loss: 0.2045 - val_loss: 0.2560\n",
      "Epoch 54/100\n",
      "32965/32965 [==============================] - 2s 47us/sample - loss: 0.2045 - val_loss: 0.2840\n",
      "Epoch 55/100\n",
      "32965/32965 [==============================] - 2s 47us/sample - loss: 0.2039 - val_loss: 0.2504\n",
      "Epoch 56/100\n",
      "32965/32965 [==============================] - 2s 48us/sample - loss: 0.2029 - val_loss: 0.2658\n",
      "Epoch 57/100\n",
      "32965/32965 [==============================] - 2s 47us/sample - loss: 0.2026 - val_loss: 0.2612\n",
      "Epoch 58/100\n",
      "32965/32965 [==============================] - 2s 49us/sample - loss: 0.2024 - val_loss: 0.2745\n",
      "Epoch 59/100\n",
      "32965/32965 [==============================] - 2s 51us/sample - loss: 0.2020 - val_loss: 0.3300\n",
      "Epoch 60/100\n",
      "32965/32965 [==============================] - 2s 51us/sample - loss: 0.2011 - val_loss: 0.3097\n",
      "Epoch 61/100\n",
      "32965/32965 [==============================] - 2s 50us/sample - loss: 0.2016 - val_loss: 0.3137\n",
      "Epoch 62/100\n",
      "32965/32965 [==============================] - 2s 53us/sample - loss: 0.2003 - val_loss: 0.2815\n",
      "Epoch 63/100\n",
      "32965/32965 [==============================] - 2s 51us/sample - loss: 0.2009 - val_loss: 0.2614\n",
      "Epoch 64/100\n",
      "32965/32965 [==============================] - 2s 50us/sample - loss: 0.1997 - val_loss: 0.2755\n",
      "Epoch 65/100\n",
      "32965/32965 [==============================] - 2s 51us/sample - loss: 0.1988 - val_loss: 0.3231\n",
      "Epoch 66/100\n",
      "32965/32965 [==============================] - 2s 51us/sample - loss: 0.1989 - val_loss: 0.2833\n",
      "Epoch 67/100\n",
      "32965/32965 [==============================] - 2s 52us/sample - loss: 0.1987 - val_loss: 0.3083\n",
      "Epoch 68/100\n",
      "32965/32965 [==============================] - 2s 66us/sample - loss: 0.1988 - val_loss: 0.3130\n",
      "Epoch 69/100\n",
      "32965/32965 [==============================] - 2s 53us/sample - loss: 0.1972 - val_loss: 0.3068\n",
      "Epoch 70/100\n",
      "32965/32965 [==============================] - 2s 52us/sample - loss: 0.1978 - val_loss: 0.3533\n",
      "Epoch 71/100\n",
      "32965/32965 [==============================] - 2s 52us/sample - loss: 0.1976 - val_loss: 0.2731\n",
      "Epoch 72/100\n",
      "32965/32965 [==============================] - 2s 52us/sample - loss: 0.1971 - val_loss: 0.2990\n",
      "Epoch 73/100\n",
      "32965/32965 [==============================] - 2s 52us/sample - loss: 0.1981 - val_loss: 0.3003\n",
      "Completed monte carlo training loop 1 of 1\n"
     ]
    }
   ],
   "source": [
    "cluster_training(inputs=inputs_SL2P, \n",
    "                 outputs=outputs,\n",
    "                 cluster_labels=cluster_labels_SL2P,\n",
    "                 keras_model=SL2P_model, batch_size=32, epochs_per_loop=100, verbosity=1, callbacks=[stop_on_val_loss], num_batches=100, validation_split=0.2, monte_carlo_loops=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 41472 samples\n",
      "Epoch 1/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.3975\n",
      "Epoch 2/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.3821\n",
      "Epoch 3/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.3700\n",
      "Epoch 4/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.3585\n",
      "Epoch 5/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.3431\n",
      "Epoch 6/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.3233\n",
      "Epoch 7/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.3059\n",
      "Epoch 8/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.2887\n",
      "Epoch 9/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.2755\n",
      "Epoch 10/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.2646\n",
      "Epoch 11/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.2550\n",
      "Epoch 12/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.2482\n",
      "Epoch 13/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.2417\n",
      "Epoch 14/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.2370\n",
      "Epoch 15/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.2327\n",
      "Epoch 16/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.2286\n",
      "Epoch 17/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.2259\n",
      "Epoch 18/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.2229\n",
      "Epoch 19/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.2209\n",
      "Epoch 20/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.2184\n",
      "Epoch 21/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.2165\n",
      "Epoch 22/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.2157\n",
      "Epoch 23/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.2138\n",
      "Epoch 24/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.2135\n",
      "Epoch 25/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.2123\n",
      "Epoch 26/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.21130s - \n",
      "Epoch 27/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.2101\n",
      "Epoch 28/1000\n",
      "41472/41472 [==============================] - 2s 43us/sample - loss: 0.2091\n",
      "Epoch 29/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.2081\n",
      "Epoch 30/1000\n",
      "41472/41472 [==============================] - ETA: 0s - loss: 0.207 - 2s 42us/sample - loss: 0.2071\n",
      "Epoch 31/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.2071\n",
      "Epoch 32/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.2064\n",
      "Epoch 33/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.2058\n",
      "Epoch 34/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.2050\n",
      "Epoch 35/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.2040\n",
      "Epoch 36/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.2035\n",
      "Epoch 37/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.2043\n",
      "Epoch 38/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.2031\n",
      "Epoch 39/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.20300s - loss:\n",
      "Epoch 40/1000\n",
      "41472/41472 [==============================] - 2s 47us/sample - loss: 0.2021\n",
      "Epoch 41/1000\n",
      "41472/41472 [==============================] - 2s 46us/sample - loss: 0.2009\n",
      "Epoch 42/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.2004\n",
      "Epoch 43/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.2013\n",
      "Epoch 44/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.1986\n",
      "Epoch 45/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.1989\n",
      "Epoch 46/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.1985\n",
      "Epoch 47/1000\n",
      "41472/41472 [==============================] - 2s 49us/sample - loss: 0.1976\n",
      "Epoch 48/1000\n",
      "41472/41472 [==============================] - 2s 57us/sample - loss: 0.1989\n",
      "Epoch 49/1000\n",
      "41472/41472 [==============================] - 2s 57us/sample - loss: 0.1982\n",
      "Epoch 50/1000\n",
      "41472/41472 [==============================] - 2s 43us/sample - loss: 0.1980\n",
      "Epoch 51/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.1978\n",
      "Epoch 52/1000\n",
      "41472/41472 [==============================] - 2s 44us/sample - loss: 0.1966\n",
      "Epoch 53/1000\n",
      "41472/41472 [==============================] - 2s 43us/sample - loss: 0.1964\n",
      "Epoch 54/1000\n",
      "41472/41472 [==============================] - 2s 44us/sample - loss: 0.1977\n",
      "Epoch 55/1000\n",
      "41472/41472 [==============================] - 2s 45us/sample - loss: 0.1956\n",
      "Epoch 56/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.1964\n",
      "Epoch 57/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.1950\n",
      "Epoch 58/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1948\n",
      "Epoch 59/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.1955\n",
      "Epoch 60/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.1949\n",
      "Epoch 61/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1949\n",
      "Epoch 62/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1942\n",
      "Epoch 63/1000\n",
      "41472/41472 [==============================] - 1s 34us/sample - loss: 0.1942\n",
      "Epoch 64/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.1947\n",
      "Epoch 65/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1942\n",
      "Epoch 66/1000\n",
      "41472/41472 [==============================] - 1s 34us/sample - loss: 0.1943\n",
      "Epoch 67/1000\n",
      "41472/41472 [==============================] - 1s 34us/sample - loss: 0.1934\n",
      "Epoch 68/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.1933\n",
      "Epoch 69/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.1931\n",
      "Epoch 70/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.1937\n",
      "Epoch 71/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.1934\n",
      "Epoch 72/1000\n",
      "41472/41472 [==============================] - 1s 34us/sample - loss: 0.1924\n",
      "Epoch 73/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.1933\n",
      "Epoch 74/1000\n",
      "41472/41472 [==============================] - 1s 34us/sample - loss: 0.1925\n",
      "Epoch 75/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1927\n",
      "Epoch 76/1000\n",
      "41472/41472 [==============================] - 2s 57us/sample - loss: 0.1923\n",
      "Epoch 77/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.1922\n",
      "Epoch 78/1000\n",
      "41472/41472 [==============================] - 2s 47us/sample - loss: 0.1929\n",
      "Epoch 79/1000\n",
      "41472/41472 [==============================] - 2s 52us/sample - loss: 0.1924\n",
      "Epoch 80/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.1932\n",
      "Epoch 81/1000\n",
      "41472/41472 [==============================] - 2s 44us/sample - loss: 0.1913\n",
      "Epoch 82/1000\n",
      "41472/41472 [==============================] - 2s 43us/sample - loss: 0.1920\n",
      "Epoch 83/1000\n",
      "41472/41472 [==============================] - 2s 43us/sample - loss: 0.1921\n",
      "Epoch 84/1000\n",
      "41472/41472 [==============================] - 2s 50us/sample - loss: 0.1918\n",
      "Epoch 85/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1917\n",
      "Epoch 86/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.1920\n",
      "Epoch 87/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.1925\n",
      "Epoch 88/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.1920\n",
      "Epoch 89/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.19030s\n",
      "Epoch 90/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1909\n",
      "Epoch 91/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1917\n",
      "Epoch 92/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.1918\n",
      "Epoch 93/1000\n",
      "41472/41472 [==============================] - 2s 47us/sample - loss: 0.1912\n",
      "Epoch 94/1000\n",
      "41472/41472 [==============================] - 3s 61us/sample - loss: 0.1906\n",
      "Epoch 95/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.1909\n",
      "Epoch 96/1000\n",
      "41472/41472 [==============================] - 2s 46us/sample - loss: 0.1908\n",
      "Epoch 97/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.1910\n",
      "Epoch 98/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.1907\n",
      "Epoch 99/1000\n",
      "41472/41472 [==============================] - 2s 45us/sample - loss: 0.1910\n",
      "Epoch 100/1000\n",
      "41472/41472 [==============================] - 2s 43us/sample - loss: 0.1909\n",
      "Epoch 101/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.1910\n",
      "Epoch 102/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.1913\n",
      "Epoch 103/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.1907\n",
      "Epoch 104/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1910\n",
      "Epoch 105/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1906\n",
      "Epoch 106/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.1910\n",
      "Epoch 107/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.1902\n",
      "Epoch 108/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.1896\n",
      "Epoch 109/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1907\n",
      "Epoch 110/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.1907\n",
      "Epoch 111/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1900\n",
      "Epoch 112/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.1907\n",
      "Epoch 113/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1898\n",
      "Epoch 114/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.1901\n",
      "Epoch 115/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.1899\n",
      "Epoch 116/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.1899\n",
      "Epoch 117/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.1901\n",
      "Epoch 118/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.1901\n",
      "Epoch 119/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1902\n",
      "Epoch 120/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.1900\n",
      "Epoch 121/1000\n",
      "41472/41472 [==============================] - 2s 56us/sample - loss: 0.1892\n",
      "Epoch 122/1000\n",
      "41472/41472 [==============================] - 2s 43us/sample - loss: 0.1902\n",
      "Epoch 123/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.1906\n",
      "Epoch 124/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.1889\n",
      "Epoch 125/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.1899\n",
      "Epoch 126/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.1899\n",
      "Epoch 127/1000\n",
      "41472/41472 [==============================] - 2s 43us/sample - loss: 0.1899\n",
      "Epoch 128/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.1890\n",
      "Epoch 129/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.1891\n",
      "Epoch 130/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.1892\n",
      "Epoch 131/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.1889\n",
      "Epoch 132/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.1899\n",
      "Epoch 133/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1897\n",
      "Epoch 134/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1890\n",
      "Epoch 135/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.1891\n",
      "Epoch 136/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.1892\n",
      "Epoch 137/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.1897\n",
      "Epoch 138/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.1896\n",
      "Epoch 139/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.1897\n",
      "Epoch 140/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.1891\n",
      "Epoch 141/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.1886\n",
      "Epoch 142/1000\n",
      "41472/41472 [==============================] - 2s 54us/sample - loss: 0.1896\n",
      "Epoch 143/1000\n",
      "41472/41472 [==============================] - 2s 43us/sample - loss: 0.1900\n",
      "Epoch 144/1000\n",
      "41472/41472 [==============================] - 2s 53us/sample - loss: 0.1891\n",
      "Epoch 145/1000\n",
      "41472/41472 [==============================] - 2s 44us/sample - loss: 0.1890\n",
      "Epoch 146/1000\n",
      "41472/41472 [==============================] - 3s 61us/sample - loss: 0.1885\n",
      "Epoch 147/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.1884\n",
      "Epoch 148/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.1893\n",
      "Epoch 149/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.1880\n",
      "Epoch 150/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.1887\n",
      "Epoch 151/1000\n",
      "41472/41472 [==============================] - 2s 56us/sample - loss: 0.1890\n",
      "Epoch 152/1000\n",
      "41472/41472 [==============================] - 2s 57us/sample - loss: 0.1882\n",
      "Epoch 153/1000\n",
      "41472/41472 [==============================] - 2s 44us/sample - loss: 0.1881\n",
      "Epoch 154/1000\n",
      "41472/41472 [==============================] - 2s 57us/sample - loss: 0.1882\n",
      "Epoch 155/1000\n",
      "41472/41472 [==============================] - 2s 47us/sample - loss: 0.1887\n",
      "Epoch 156/1000\n",
      "41472/41472 [==============================] - 2s 53us/sample - loss: 0.1881\n",
      "Epoch 157/1000\n",
      "41472/41472 [==============================] - 2s 54us/sample - loss: 0.1881\n",
      "Epoch 158/1000\n",
      "41472/41472 [==============================] - 3s 63us/sample - loss: 0.1880\n",
      "Epoch 159/1000\n",
      "41472/41472 [==============================] - 3s 69us/sample - loss: 0.1882\n",
      "Epoch 160/1000\n",
      "41472/41472 [==============================] - 2s 56us/sample - loss: 0.1869\n",
      "Epoch 161/1000\n",
      "41472/41472 [==============================] - 2s 54us/sample - loss: 0.1873\n",
      "Epoch 162/1000\n",
      "41472/41472 [==============================] - 2s 58us/sample - loss: 0.1877\n",
      "Epoch 163/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.1879\n",
      "Epoch 164/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.1885\n",
      "Epoch 165/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.1873\n",
      "Epoch 166/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.1880\n",
      "Epoch 167/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.1884\n",
      "Epoch 168/1000\n",
      "41472/41472 [==============================] - 3s 70us/sample - loss: 0.1880\n",
      "Epoch 169/1000\n",
      "41472/41472 [==============================] - 2s 45us/sample - loss: 0.1872\n",
      "Epoch 170/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.1878\n",
      "Epoch 171/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.1874\n",
      "Epoch 172/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.1872\n",
      "Epoch 173/1000\n",
      "41472/41472 [==============================] - 2s 48us/sample - loss: 0.1876\n",
      "Epoch 174/1000\n",
      "41472/41472 [==============================] - 3s 75us/sample - loss: 0.1871\n",
      "Epoch 175/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.1878\n",
      "Epoch 176/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.1870\n",
      "Epoch 177/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.1874\n",
      "Epoch 178/1000\n",
      "41472/41472 [==============================] - 2s 49us/sample - loss: 0.1877\n",
      "Epoch 179/1000\n",
      "41472/41472 [==============================] - 2s 50us/sample - loss: 0.1871\n",
      "Epoch 180/1000\n",
      "41472/41472 [==============================] - 2s 59us/sample - loss: 0.1877\n",
      "Epoch 181/1000\n",
      "41472/41472 [==============================] - 3s 61us/sample - loss: 0.1868\n",
      "Epoch 182/1000\n",
      "41472/41472 [==============================] - 2s 57us/sample - loss: 0.1871\n",
      "Epoch 183/1000\n",
      "41472/41472 [==============================] - 2s 53us/sample - loss: 0.1874\n",
      "Epoch 184/1000\n",
      "41472/41472 [==============================] - 2s 47us/sample - loss: 0.1870\n",
      "Epoch 185/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.1871\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x1fde4ee8448>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "SL2P_model.fit(x=scaled_inputs_SL2P, y=outputs.values, batch_size=32, epochs=1000, verbose=1, callbacks=[stop_on_loss], shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Predicting our own input data to evaluate the performance (for now)\n",
    "SL2P_predictions = SL2P_model.predict(scaled_inputs_SL2P).flatten()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_prediction(true_values=outputs, predicted_values=SL2P_predictions, ideal_predictor_range=[0,14], description=\"SL2P Neural Net LAI prediction\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "band_measurement_error = pandas.Series([0.025]*inputs_SL2P.shape[1], name=\"error\")\n",
    "band_measurement_error.index = features_SL2P\n",
    "\n",
    "RMSE_SL2P = pandas.Series(data=cluster_RMSE(inputs=inputs_SL2P, outputs=outputs, predictions=pandas.Series(SL2P_predictions, name=\"SL2P\"), band_errors=band_measurement_error), name=\"RMSE\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Using Keras to create a sequential model neural network which only has simple dense layers of the specified number of nodes\n",
    "SL2P_error_model = tensorflow.keras.models.Sequential([\n",
    "    tensorflow.keras.layers.Dense(11, input_shape = [inputs_SL2P.shape[1]]),\n",
    "    tensorflow.keras.layers.Dense(5, activation='tansig'),\n",
    "    tensorflow.keras.layers.Dense(1)\n",
    "])\n",
    "# Compiling the model to minimize the mean squared error loss function and use the NADAM optimizer\n",
    "SL2P_error_model.compile(\n",
    "    optimizer=tensorflow.keras.optimizers.Nadam(),\n",
    "    loss='mse')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 32718 samples, validate on 8754 samples\n",
      "Epoch 1/100\n",
      "32718/32718 [==============================] - 3s 88us/sample - loss: 0.0684 - val_loss: 0.0413\n",
      "Epoch 2/100\n",
      "32718/32718 [==============================] - 2s 46us/sample - loss: 0.0375 - val_loss: 0.0399\n",
      "Epoch 3/100\n",
      "32718/32718 [==============================] - 2s 53us/sample - loss: 0.0370 - val_loss: 0.0397\n",
      "Epoch 4/100\n",
      "32718/32718 [==============================] - 2s 51us/sample - loss: 0.0367 - val_loss: 0.0388\n",
      "Epoch 5/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0364 - val_loss: 0.0380\n",
      "Epoch 6/100\n",
      "32718/32718 [==============================] - 2s 54us/sample - loss: 0.0362 - val_loss: 0.0388\n",
      "Epoch 7/100\n",
      "32718/32718 [==============================] - 2s 58us/sample - loss: 0.0360 - val_loss: 0.0380\n",
      "Epoch 8/100\n",
      "32718/32718 [==============================] - 2s 54us/sample - loss: 0.0358 - val_loss: 0.0377\n",
      "Epoch 9/100\n",
      "32718/32718 [==============================] - 2s 46us/sample - loss: 0.0355 - val_loss: 0.0372\n",
      "Epoch 10/100\n",
      "32718/32718 [==============================] - 2s 51us/sample - loss: 0.0345 - val_loss: 0.0378\n",
      "Epoch 11/100\n",
      "32718/32718 [==============================] - 1s 45us/sample - loss: 0.0342 - val_loss: 0.0370\n",
      "Epoch 12/100\n",
      "32718/32718 [==============================] - 2s 50us/sample - loss: 0.0341 - val_loss: 0.0362\n",
      "Epoch 13/100\n",
      "32718/32718 [==============================] - 2s 50us/sample - loss: 0.0341 - val_loss: 0.0358\n",
      "Epoch 14/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0341 - val_loss: 0.0364\n",
      "Epoch 15/100\n",
      "32718/32718 [==============================] - 2s 47us/sample - loss: 0.0340 - val_loss: 0.0365\n",
      "Epoch 16/100\n",
      "32718/32718 [==============================] - 2s 51us/sample - loss: 0.0340 - val_loss: 0.0373\n",
      "Epoch 17/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0339 - val_loss: 0.0364\n",
      "Epoch 18/100\n",
      "32718/32718 [==============================] - 1s 44us/sample - loss: 0.0338 - val_loss: 0.0360\n",
      "Epoch 19/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0339 - val_loss: 0.0370\n",
      "Epoch 20/100\n",
      "32718/32718 [==============================] - 2s 49us/sample - loss: 0.0338 - val_loss: 0.0358\n",
      "Epoch 21/100\n",
      "32718/32718 [==============================] - 2s 47us/sample - loss: 0.0338 - val_loss: 0.0365\n",
      "Epoch 22/100\n",
      "32718/32718 [==============================] - 1s 46us/sample - loss: 0.0337 - val_loss: 0.0372\n",
      "Epoch 23/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0336 - val_loss: 0.0359\n",
      "Epoch 24/100\n",
      "32718/32718 [==============================] - 2s 51us/sample - loss: 0.0337 - val_loss: 0.0365\n",
      "Epoch 25/100\n",
      "32718/32718 [==============================] - 1s 45us/sample - loss: 0.0337 - val_loss: 0.0359\n",
      "Epoch 26/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0336 - val_loss: 0.0373\n",
      "Epoch 27/100\n",
      "32718/32718 [==============================] - 2s 50us/sample - loss: 0.0336 - val_loss: 0.0363\n",
      "Epoch 28/100\n",
      "32718/32718 [==============================] - 2s 46us/sample - loss: 0.0336 - val_loss: 0.0369\n",
      "Epoch 29/100\n",
      "32718/32718 [==============================] - 1s 45us/sample - loss: 0.0336 - val_loss: 0.0370\n",
      "Epoch 30/100\n",
      "32718/32718 [==============================] - 2s 51us/sample - loss: 0.0335 - val_loss: 0.0354\n",
      "Epoch 31/100\n",
      "32718/32718 [==============================] - 2s 49us/sample - loss: 0.0335 - val_loss: 0.0365\n",
      "Epoch 32/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0335 - val_loss: 0.0361\n",
      "Epoch 33/100\n",
      "32718/32718 [==============================] - 2s 63us/sample - loss: 0.0335 - val_loss: 0.0364\n",
      "Epoch 34/100\n",
      "32718/32718 [==============================] - 2s 47us/sample - loss: 0.0335 - val_loss: 0.0358\n",
      "Epoch 35/100\n",
      "32718/32718 [==============================] - 1s 45us/sample - loss: 0.0334 - val_loss: 0.0359\n",
      "Epoch 36/100\n",
      "32718/32718 [==============================] - 1s 45us/sample - loss: 0.0334 - val_loss: 0.0366\n",
      "Epoch 37/100\n",
      "32718/32718 [==============================] - 2s 47us/sample - loss: 0.0334 - val_loss: 0.0367\n",
      "Epoch 38/100\n",
      "32718/32718 [==============================] - 2s 47us/sample - loss: 0.0334 - val_loss: 0.0369\n",
      "Epoch 39/100\n",
      "32718/32718 [==============================] - 2s 46us/sample - loss: 0.0334 - val_loss: 0.0354\n",
      "Epoch 40/100\n",
      "32718/32718 [==============================] - 2s 46us/sample - loss: 0.0333 - val_loss: 0.0356\n",
      "Epoch 41/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0333 - val_loss: 0.0361\n",
      "Epoch 42/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0333 - val_loss: 0.0364\n",
      "Epoch 43/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0333 - val_loss: 0.0358\n",
      "Epoch 44/100\n",
      "32718/32718 [==============================] - 2s 54us/sample - loss: 0.0333 - val_loss: 0.0355\n",
      "Epoch 45/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0332 - val_loss: 0.0367\n",
      "Epoch 46/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0332 - val_loss: 0.0357\n",
      "Epoch 47/100\n",
      "32718/32718 [==============================] - 2s 47us/sample - loss: 0.0332 - val_loss: 0.0353\n",
      "Epoch 48/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0332 - val_loss: 0.0359\n",
      "Epoch 49/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0332 - val_loss: 0.0368\n",
      "Epoch 50/100\n",
      "32718/32718 [==============================] - 2s 49us/sample - loss: 0.0332 - val_loss: 0.0358\n",
      "Epoch 51/100\n",
      "32718/32718 [==============================] - 2s 55us/sample - loss: 0.0332 - val_loss: 0.0364\n",
      "Epoch 52/100\n",
      "32718/32718 [==============================] - 2s 58us/sample - loss: 0.0331 - val_loss: 0.0354\n",
      "Epoch 53/100\n",
      "32718/32718 [==============================] - 2s 52us/sample - loss: 0.0332 - val_loss: 0.0357\n",
      "Epoch 54/100\n",
      "32718/32718 [==============================] - 2s 52us/sample - loss: 0.0331 - val_loss: 0.0367\n",
      "Epoch 55/100\n",
      "32718/32718 [==============================] - 2s 52us/sample - loss: 0.0331 - val_loss: 0.0357\n",
      "Epoch 56/100\n",
      "32718/32718 [==============================] - 2s 51us/sample - loss: 0.0331 - val_loss: 0.0358\n",
      "Epoch 57/100\n",
      "32718/32718 [==============================] - 2s 52us/sample - loss: 0.0331 - val_loss: 0.0362\n",
      "Epoch 58/100\n",
      "32718/32718 [==============================] - 2s 53us/sample - loss: 0.0331 - val_loss: 0.0351\n",
      "Epoch 59/100\n",
      "32718/32718 [==============================] - 2s 56us/sample - loss: 0.0330 - val_loss: 0.0361\n",
      "Epoch 60/100\n",
      "32718/32718 [==============================] - 2s 71us/sample - loss: 0.0331 - val_loss: 0.0361\n",
      "Epoch 61/100\n",
      "32718/32718 [==============================] - 2s 63us/sample - loss: 0.0330 - val_loss: 0.0361\n",
      "Epoch 62/100\n",
      "32718/32718 [==============================] - 3s 92us/sample - loss: 0.0330 - val_loss: 0.0354\n",
      "Epoch 63/100\n",
      "32718/32718 [==============================] - 3s 84us/sample - loss: 0.0330 - val_loss: 0.0358\n",
      "Epoch 64/100\n",
      "32718/32718 [==============================] - 5s 150us/sample - loss: 0.0330 - val_loss: 0.0358\n",
      "Epoch 65/100\n",
      "32718/32718 [==============================] - 3s 87us/sample - loss: 0.0330 - val_loss: 0.0354\n",
      "Epoch 66/100\n",
      "32718/32718 [==============================] - 2s 55us/sample - loss: 0.0330 - val_loss: 0.0354\n",
      "Epoch 67/100\n",
      "32718/32718 [==============================] - 2s 53us/sample - loss: 0.0330 - val_loss: 0.0346\n",
      "Epoch 68/100\n",
      "32718/32718 [==============================] - 2s 56us/sample - loss: 0.0330 - val_loss: 0.0370\n",
      "Epoch 69/100\n",
      "32718/32718 [==============================] - 2s 55us/sample - loss: 0.0330 - val_loss: 0.0352\n",
      "Epoch 70/100\n",
      "32718/32718 [==============================] - 2s 53us/sample - loss: 0.0329 - val_loss: 0.0364\n",
      "Epoch 71/100\n",
      "32718/32718 [==============================] - 2s 53us/sample - loss: 0.0329 - val_loss: 0.0366\n",
      "Epoch 72/100\n",
      "32718/32718 [==============================] - 2s 54us/sample - loss: 0.0330 - val_loss: 0.0361\n",
      "Epoch 73/100\n",
      "32718/32718 [==============================] - 2s 48us/sample - loss: 0.0329 - val_loss: 0.0361\n",
      "Epoch 74/100\n",
      "32718/32718 [==============================] - 2s 49us/sample - loss: 0.0329 - val_loss: 0.0357\n",
      "Epoch 75/100\n",
      "32718/32718 [==============================] - 2s 50us/sample - loss: 0.0329 - val_loss: 0.0357\n",
      "Epoch 76/100\n",
      "32718/32718 [==============================] - 2s 51us/sample - loss: 0.0329 - val_loss: 0.0368\n",
      "Epoch 77/100\n",
      "32718/32718 [==============================] - 2s 49us/sample - loss: 0.0329 - val_loss: 0.0356\n",
      "Epoch 78/100\n",
      "32718/32718 [==============================] - 2s 59us/sample - loss: 0.0329 - val_loss: 0.0370\n",
      "Epoch 79/100\n",
      "32718/32718 [==============================] - 3s 82us/sample - loss: 0.0329 - val_loss: 0.0354\n",
      "Epoch 80/100\n",
      "32718/32718 [==============================] - 2s 54us/sample - loss: 0.0328 - val_loss: 0.0358\n",
      "Epoch 81/100\n",
      "32718/32718 [==============================] - 2s 55us/sample - loss: 0.0329 - val_loss: 0.0353\n",
      "Epoch 82/100\n",
      "32718/32718 [==============================] - 2s 59us/sample - loss: 0.0328 - val_loss: 0.0364\n",
      "Epoch 83/100\n",
      "32718/32718 [==============================] - 2s 56us/sample - loss: 0.0328 - val_loss: 0.0350\n",
      "Epoch 84/100\n",
      "32718/32718 [==============================] - 2s 52us/sample - loss: 0.0328 - val_loss: 0.0361\n",
      "Epoch 85/100\n",
      "32718/32718 [==============================] - 2s 53us/sample - loss: 0.0328 - val_loss: 0.0364\n",
      "Epoch 86/100\n",
      "32718/32718 [==============================] - 2s 53us/sample - loss: 0.0328 - val_loss: 0.0363\n",
      "Epoch 87/100\n",
      "32718/32718 [==============================] - 2s 49us/sample - loss: 0.0328 - val_loss: 0.0365\n",
      "Epoch 88/100\n",
      "32718/32718 [==============================] - 2s 50us/sample - loss: 0.0328 - val_loss: 0.0357\n",
      "Epoch 89/100\n",
      "32718/32718 [==============================] - 1s 44us/sample - loss: 0.0328 - val_loss: 0.0352\n",
      "Epoch 90/100\n",
      "32718/32718 [==============================] - 1s 45us/sample - loss: 0.0328 - val_loss: 0.0377\n",
      "Epoch 91/100\n",
      "32718/32718 [==============================] - 1s 44us/sample - loss: 0.0327 - val_loss: 0.0355\n",
      "Epoch 92/100\n",
      "32718/32718 [==============================] - 2s 46us/sample - loss: 0.0328 - val_loss: 0.0370\n",
      "Completed monte carlo training loop 1 of 1\n"
     ]
    }
   ],
   "source": [
    "cluster_training(inputs=inputs_SL2P, outputs=RMSE_SL2P, keras_model=SL2P_error_model, callbacks=[stop_on_val_loss], monte_carlo_loops=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 41472 samples\n",
      "Epoch 1/1000\n",
      "41472/41472 [==============================] - 2s 45us/sample - loss: 0.0334\n",
      "Epoch 2/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.0333\n",
      "Epoch 3/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.0333\n",
      "Epoch 4/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.0333\n",
      "Epoch 5/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.0333\n",
      "Epoch 6/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.0332\n",
      "Epoch 7/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.0332\n",
      "Epoch 8/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.0332\n",
      "Epoch 9/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.0332\n",
      "Epoch 10/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0332\n",
      "Epoch 11/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.0332\n",
      "Epoch 12/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0332\n",
      "Epoch 13/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.0332\n",
      "Epoch 14/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0332\n",
      "Epoch 15/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.0332\n",
      "Epoch 16/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0332\n",
      "Epoch 17/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.0331\n",
      "Epoch 18/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.0331\n",
      "Epoch 19/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.0332\n",
      "Epoch 20/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.0331\n",
      "Epoch 21/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.0331\n",
      "Epoch 22/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.0331\n",
      "Epoch 23/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0331\n",
      "Epoch 24/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.0331\n",
      "Epoch 25/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.0330\n",
      "Epoch 26/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.0330\n",
      "Epoch 27/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.0330\n",
      "Epoch 28/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.0330\n",
      "Epoch 29/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.0331\n",
      "Epoch 30/1000\n",
      "41472/41472 [==============================] - 1s 34us/sample - loss: 0.0330\n",
      "Epoch 31/1000\n",
      "41472/41472 [==============================] - 1s 34us/sample - loss: 0.0330\n",
      "Epoch 32/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.0330\n",
      "Epoch 33/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.0329\n",
      "Epoch 34/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.0330\n",
      "Epoch 35/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.0330\n",
      "Epoch 36/1000\n",
      "41472/41472 [==============================] - 2s 36us/sample - loss: 0.0329\n",
      "Epoch 37/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0329\n",
      "Epoch 38/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0329\n",
      "Epoch 39/1000\n",
      "41472/41472 [==============================] - 2s 37us/sample - loss: 0.0329\n",
      "Epoch 40/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0329\n",
      "Epoch 41/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0329\n",
      "Epoch 42/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0329\n",
      "Epoch 43/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0329\n",
      "Epoch 44/1000\n",
      "41472/41472 [==============================] - 1s 35us/sample - loss: 0.0329\n",
      "Epoch 45/1000\n",
      "41472/41472 [==============================] - 1s 36us/sample - loss: 0.0329\n",
      "Epoch 46/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.0328\n",
      "Epoch 47/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.0328\n",
      "Epoch 48/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.0329\n",
      "Epoch 49/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0328\n",
      "Epoch 50/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0329\n",
      "Epoch 51/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.0328\n",
      "Epoch 52/1000\n",
      "41472/41472 [==============================] - 2s 38us/sample - loss: 0.0328\n",
      "Epoch 53/1000\n",
      "41472/41472 [==============================] - 2s 50us/sample - loss: 0.0328\n",
      "Epoch 54/1000\n",
      "41472/41472 [==============================] - 2s 46us/sample - loss: 0.0328\n",
      "Epoch 55/1000\n",
      "41472/41472 [==============================] - 2s 40us/sample - loss: 0.0328\n",
      "Epoch 56/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.0328\n",
      "Epoch 57/1000\n",
      "41472/41472 [==============================] - 2s 39us/sample - loss: 0.0328\n",
      "Epoch 58/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.0328\n",
      "Epoch 59/1000\n",
      "41472/41472 [==============================] - 2s 44us/sample - loss: 0.0328\n",
      "Epoch 60/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.0328\n",
      "Epoch 61/1000\n",
      "41472/41472 [==============================] - 2s 44us/sample - loss: 0.0328\n",
      "Epoch 62/1000\n",
      "41472/41472 [==============================] - 2s 50us/sample - loss: 0.0328\n",
      "Epoch 63/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.0328\n",
      "Epoch 64/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.0328\n",
      "Epoch 65/1000\n",
      "41472/41472 [==============================] - 2s 44us/sample - loss: 0.0327\n",
      "Epoch 66/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.0328\n",
      "Epoch 67/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.0328\n",
      "Epoch 68/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.0327\n",
      "Epoch 69/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.0328\n",
      "Epoch 70/1000\n",
      "41472/41472 [==============================] - 2s 41us/sample - loss: 0.0327\n",
      "Epoch 71/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.0327\n",
      "Epoch 72/1000\n",
      "41472/41472 [==============================] - 2s 42us/sample - loss: 0.0328\n",
      "Epoch 73/1000\n",
      "41472/41472 [==============================] - 2s 46us/sample - loss: 0.0327\n",
      "Epoch 74/1000\n",
      "41472/41472 [==============================] - 2s 45us/sample - loss: 0.0327\n",
      "Epoch 75/1000\n",
      "41472/41472 [==============================] - 2s 46us/sample - loss: 0.0328\n",
      "Epoch 76/1000\n",
      "41472/41472 [==============================] - 2s 45us/sample - loss: 0.0327\n",
      "Epoch 77/1000\n",
      "41472/41472 [==============================] - 2s 45us/sample - loss: 0.0327\n",
      "Epoch 78/1000\n",
      "41472/41472 [==============================] - 2s 47us/sample - loss: 0.0327\n",
      "Epoch 79/1000\n",
      "41472/41472 [==============================] - 2s 46us/sample - loss: 0.0327\n",
      "Epoch 80/1000\n",
      "41472/41472 [==============================] - 2s 44us/sample - loss: 0.0328\n",
      "Epoch 81/1000\n",
      "41472/41472 [==============================] - 2s 47us/sample - loss: 0.0327\n",
      "Epoch 82/1000\n",
      "41472/41472 [==============================] - 2s 45us/sample - loss: 0.0327\n",
      "Epoch 83/1000\n",
      "41472/41472 [==============================] - 2s 52us/sample - loss: 0.0326\n",
      "Epoch 84/1000\n",
      "41472/41472 [==============================] - 2s 48us/sample - loss: 0.0327\n",
      "Epoch 85/1000\n",
      "41472/41472 [==============================] - 2s 49us/sample - loss: 0.0327\n",
      "Epoch 86/1000\n",
      "41472/41472 [==============================] - 2s 56us/sample - loss: 0.0327\n",
      "Epoch 87/1000\n",
      "41472/41472 [==============================] - 2s 47us/sample - loss: 0.0326\n",
      "Epoch 88/1000\n",
      "41472/41472 [==============================] - 2s 47us/sample - loss: 0.0326\n",
      "Epoch 89/1000\n",
      "41472/41472 [==============================] - 2s 46us/sample - loss: 0.0326\n",
      "Epoch 90/1000\n",
      "41472/41472 [==============================] - 2s 49us/sample - loss: 0.0326\n",
      "Epoch 91/1000\n",
      "41472/41472 [==============================] - 2s 48us/sample - loss: 0.0326\n",
      "Epoch 92/1000\n",
      "41472/41472 [==============================] - 2s 46us/sample - loss: 0.0327\n",
      "Epoch 93/1000\n",
      "41472/41472 [==============================] - 2s 49us/sample - loss: 0.0327\n",
      "Epoch 94/1000\n",
      "41472/41472 [==============================] - 2s 50us/sample - loss: 0.0326\n",
      "Epoch 95/1000\n",
      "41472/41472 [==============================] - 2s 44us/sample - loss: 0.0326\n",
      "Epoch 96/1000\n",
      "41472/41472 [==============================] - 2s 52us/sample - loss: 0.0327\n",
      "Epoch 97/1000\n",
      "41472/41472 [==============================] - 2s 46us/sample - loss: 0.0326\n",
      "Epoch 98/1000\n",
      "41472/41472 [==============================] - 2s 44us/sample - loss: 0.0326\n",
      "Epoch 99/1000\n",
      "41472/41472 [==============================] - 2s 46us/sample - loss: 0.0326\n",
      "Epoch 100/1000\n",
      "41472/41472 [==============================] - 2s 56us/sample - loss: 0.0325\n",
      "Epoch 101/1000\n",
      "41472/41472 [==============================] - 2s 49us/sample - loss: 0.0326\n",
      "Epoch 102/1000\n",
      "41472/41472 [==============================] - 2s 48us/sample - loss: 0.0326\n",
      "Epoch 103/1000\n",
      "41472/41472 [==============================] - 2s 46us/sample - loss: 0.0326\n",
      "Epoch 104/1000\n",
      "41472/41472 [==============================] - 2s 45us/sample - loss: 0.0326\n",
      "Epoch 105/1000\n",
      "41472/41472 [==============================] - 2s 45us/sample - loss: 0.0325\n",
      "Epoch 106/1000\n",
      "41472/41472 [==============================] - 2s 50us/sample - loss: 0.0326\n",
      "Epoch 107/1000\n",
      "41472/41472 [==============================] - 2s 49us/sample - loss: 0.0326\n",
      "Epoch 108/1000\n",
      "41472/41472 [==============================] - 2s 55us/sample - loss: 0.0326\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x1fde4736288>"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "SL2P_error_model.fit(x=scaled_inputs_SL2P, y=RMSE_SL2P, batch_size=32, epochs=1000, verbose=1, callbacks=[stop_on_loss], shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Predicting our own input data to evaluate the performance (for now)\n",
    "SL2P_error_predictions = SL2P_error_model.predict(scaled_inputs_SL2P).flatten()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_prediction(true_values=RMSE_SL2P, predicted_values=SL2P_error_predictions, ideal_predictor_range=[0,3.5], description=\"Random Forest LAI Error prediction\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Authorizes the script with Earth Engine using the credentials.json file. Ensure either ee.Authenticate() has been called in a previous\n",
    "# script or the \"$ earthengine authenticate\" command has been run in the environment to set up the credentials.json file\n",
    "# The google account you provide during the authentication flow must have received access to Google Earth Engine\n",
    "ee.Initialize()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 349,
   "metadata": {},
   "outputs": [],
   "source": [
    "def check_ee_tasks(ee_tasks: list = []):\n",
    "    for task in ee_tasks:\n",
    "        taskStatus = ee.data.getTaskStatus(task.id)[0]\n",
    "        print(taskStatus[\"description\"] + \": \" + taskStatus[\"state\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 350,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Wait loop for Earth Engine tasks to complete. Polls for the task status the specificed number of seconds until it is no longer active\n",
    "def task_wait_loop(ee_task, wait_interval):\n",
    "    print(ee.data.getTaskStatus(ee_task.id)[0][\"description\"]+\":\", end=\" \")\n",
    "    prev_task_status = ee.data.getTaskStatus(ee_task.id)[0][\"state\"]\n",
    "    print(prev_task_status, end=\" \")\n",
    "    while ee_task.active():\n",
    "        task_status = ee.data.getTaskStatus(ee_task.id)[0][\"state\"]\n",
    "        if(task_status != prev_task_status):\n",
    "            print(task_status, end=\" \")\n",
    "        prev_task_status = task_status\n",
    "        time.sleep(wait_interval)\n",
    "    print(ee.data.getTaskStatus(ee_task.id)[0][\"state\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 351,
   "metadata": {},
   "outputs": [],
   "source": [
    "default_asset_location = \"users/hemitshah\"\n",
    "def export_collection_to_gee(collection, num_images: int = 0, image_names: list = [], asset_folder: str = default_asset_location, scale: float = 20, max_pixels: int = 1e8, data_type: str = \"float\"):\n",
    "    \n",
    "    collection = ee.ImageCollection(collection)\n",
    "    \n",
    "    image_list = collection.toList(num_images)\n",
    "    task_list = []\n",
    "    \n",
    "    for i in range(num_images):\n",
    "        image = ee.Image(image_list.get(i))\n",
    "        name = image_names[i]\n",
    "        \n",
    "        typed_images = {\"double\": image.toDouble(), \"float\": image.toFloat(), \"byte\": image.toByte(), \"int\": image.toInt()}\n",
    "        export_task = ee.batch.Export.image.toAsset(image = typed_images[data_type],\n",
    "                                                      description = name,\n",
    "                                                      assetId = asset_folder+\"/\"+name,\n",
    "                                                      region = image.geometry(),\n",
    "                                                      scale = scale,\n",
    "                                                      maxPixels = max_pixels)\n",
    "        export_task.start()\n",
    "        task_list.append(export_task)\n",
    "    \n",
    "    return task_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 352,
   "metadata": {},
   "outputs": [],
   "source": [
    "def export_collection_to_drive(collection, num_images: int = 0, image_names: list = [], gdrive_folder: str = \"\", scale: float = 20, max_pixels: int = 1e8, data_type: str = \"float\"):\n",
    "    \n",
    "    collection = ee.ImageCollection(collection)\n",
    "    \n",
    "    image_list = collection.toList(num_images)\n",
    "    task_list = []\n",
    "    \n",
    "    for i in range(num_images):\n",
    "        image = ee.Image(image_list.get(i))\n",
    "        name = image_names[i]\n",
    "        \n",
    "        typed_images = {\"double\": image.toDouble(), \"float\": image.toFloat(), \"byte\": image.toByte(), \"int\": image.toInt()}\n",
    "        export_task = ee.batch.Export.image.toDrive(image = typed_images[data_type],\n",
    "                                                    description = name,\n",
    "                                                    folder = gdrive_folder,\n",
    "                                                    fileNamePrefix = name,\n",
    "                                                    region = image.geometry(),\n",
    "                                                    scale = scale,\n",
    "                                                    maxPixels = max_pixels)\n",
    "        export_task.start()\n",
    "        task_list.append(export_task)\n",
    "    \n",
    "    return task_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 353,
   "metadata": {},
   "outputs": [],
   "source": [
    "def nnet_to_ee(keras_model):\n",
    "    neural_net = ee.List([])\n",
    "    for layer in keras_model.layers:\n",
    "        layer_info = layer.get_config()\n",
    "        num_nodes = layer_info[\"units\"]\n",
    "        activation = layer_info[\"activation\"]\n",
    "        if isinstance(activation, dict):\n",
    "            activation = activation[\"config\"][\"activation\"]\n",
    "        \n",
    "        layer_weights = layer.get_weights()[0].transpose()\n",
    "        layer_bias = layer.get_weights()[1].tolist()\n",
    "        \n",
    "        nodes = [ee.ImageCollection([ee.Image(weight) for weight in node.tolist()]).toBands() for node in layer_weights]\n",
    "        node_collection = ee.ImageCollection(ee.List([ee.Image(nodes[i]).set({\"bias\": layer_bias[i]}) for i in range(num_nodes)])).set({\"activation\": activation})\n",
    "        neural_net = neural_net.add(node_collection)\n",
    "    \n",
    "    return neural_net\n",
    "\n",
    "# Defining all of the activation functions for image structures in Earth Engine\n",
    "def linear(x):\n",
    "    return ee.Image(x)\n",
    "\n",
    "def elu(x):\n",
    "    return ee.ImageCollection([x.mask(x.gte(0)), x.mask(x.lt(0)).exp().subtract(1)]).mosaic()\n",
    "\n",
    "def softplus(x):\n",
    "    return x.exp().add(1).log()\n",
    "\n",
    "def softsign(x):\n",
    "    return x.divide(x.abs().add(1))\n",
    "\n",
    "def relu(x):\n",
    "    return x.max(0.0)\n",
    "\n",
    "def tanh(x):\n",
    "    return x.multiply(2).exp().subtract(1).divide(x.multiply(2).exp().add(1))\n",
    "\n",
    "def sigmoid(x):\n",
    "    return x.exp().pow(-1).add(1).pow(-1)\n",
    "\n",
    "def tansig(x):\n",
    "    return x.multiply(-2).exp().add(1).pow(-1).multiply(2).subtract(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 354,
   "metadata": {},
   "outputs": [],
   "source": [
    "SL2P = nnet_to_ee(SL2P_model)\n",
    "SL2P_error = nnet_to_ee(SL2P_error_model)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Vegetation Indices Sources\n",
    "\n",
    "1. https://www.hindawi.com/journals/js/2017/1353691/tab1/\n",
    "2. https://www.hiphen-plant.com/blog/vegetation-indices/\n",
    "3. https://gisgeography.com/sentinel-2-bands-combinations/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 355,
   "metadata": {},
   "outputs": [],
   "source": [
    "sentinel_0001 = [\"B1\", \"B2\", \"B3\", \"B4\", \"B5\", \"B6\", \"B7\", \"B8\", \"B8A\", \"B9\", \"B11\", \"B12\"]\n",
    "sentinel_001 = [\"AOT\", \"WVP\"]\n",
    "sentinel_bands = [\"SCL\", \"TCI_R\", \"TCI_G\", \"TCI_B\", \"MSK_CLDPRB\", \"MSK_SNWPRB\", \"QA10\", \"QA20\", \"QA60\"]\n",
    "sentinel_empty = [\"SCL\", \"MSK_CLDPRB\", \"MSK_SNWPRB\", \"QA10\", \"QA20\"]\n",
    "\n",
    "# Below we define a list of strings representing the expressions for each vegetation index as a function of the bands in the input image\n",
    "# More vegetation indices can be defined, but the list CANNOT contain any two vegetation indices which are a linear combination of each\n",
    "# other or LARs will fail to select the requested number of variables\n",
    "# The formatting of the expression must be \n",
    "# \"<name of VI> = <expression with band names from inputImage_bands used as variables in the form b('<band name>')\"\n",
    "input_VI_definition =[\"GI      = b('B3')/b('B4')\",\n",
    "                      \"RVI3    = b('B4')/b('B6')\",\n",
    "                      \"SR3     = b('B5')/b('B4')\",\n",
    "                      \"GM1     = b('B6')/b('B3')\",\n",
    "                      \"GM2     = b('B6')/b('B5')\",\n",
    "                      \"SR2     = b('B7')/b('B3')\",\n",
    "                      \"PSSR    = b('B7')/b('B4')\",\n",
    "                      \"SGI     = b('B8A')/b('B4')\",\n",
    "                      \"MSI     = b('B11')/b('B7')\",\n",
    "                      \"II      = b('B11')/b('B12')\",\n",
    "                      \"GVI     = (b('B8A')/b('B3'))-1\",\n",
    "                      \"PSRI    = (b('B4')-b('B3'))/b('B6')\",\n",
    "                      \"NDVI3   = ((b('B8A')-b('B4'))/(b('B8A')))+b('B4')\",\n",
    "                      \"SR5     = 1/b('B5')\",\n",
    "                      \"SR6     = b('B4')/(b('B3')*b('B5'))\",\n",
    "                      \"SR7     = b('B8A')/(b('B3')*b('B5'))\",\n",
    "                      \"IPVI    = b('B7')/(b('B7')+b('B4'))\",\n",
    "                      \"ARI     = (1/b('B3'))-(1/b('B5'))\",\n",
    "                      \"ARI2    = b('B7')*((1/b('B3'))-(1/b('B5')))\",\n",
    "                      \"NDVI    = (b('B8A')-b('B4'))/(b('B8A')+b('B4'))\",\n",
    "                      \"GNDVI   = (b('B8A')-b('B3'))/(b('B8A')+b('B3'))\",\n",
    "                      \"NDWI    = (b('B8A')-b('B11'))/(b('B8A')+b('B11'))\",\n",
    "                      \"NDREVI  = (b('B8A')-b('B5'))/(b('B8A')+b('B5'))\",\n",
    "                      \"NDGI    = (b('B3')-b('B4'))/(b('B3')+b('B4'))\",\n",
    "                      \"NDI1    = (b('B7')-b('B5'))/(b('B7')-b('B4'))\",\n",
    "                      \"NDI2    = (b('B8A')-b('B5'))/(b('B8A')-b('B4'))\",\n",
    "                      \"RENDVI  = (b('B6')-b('B5'))/(b('B6')+b('B5'))\",\n",
    "                      \"OSAVI   = (1.16*(b('B7')-b('B4')))/(b('B7')+b('B4')+0.61)\",\n",
    "                      \"NMDI    = (b('B8A')-(b('B11')-b('B12')))/(b('B8A')+(b('B11')-b('B12')))\",\n",
    "                      \"HI      = ((b('B3')-b('B5'))/(b('B3')+b('B5')))-0.5*b('B5')\",\n",
    "                      \"GVSP    = (-0.283*b('B3') - 0.66*b('B4') + 0.577*b('B6') + 0.388*b('B8A'))/(0.433*b('B3') - 0.632*b('B4') + 0.586*b('B6') + 0.264*b('B8A'))\",\n",
    "                      \"MCARI   = ((b('B5')-b('B4'))-0.2*(b('B5')-b('B3')))*(b('B5')/b('B4'))\",\n",
    "                      \"TCARI   = 3*((b('B5')-b('B4'))-0.2*(b('B5')-b('B3'))*(b('B5')/b('B4')))\",\n",
    "                      \"EVI     = 2.5*((b('B8A')-b('B4'))/(b('B8A')+6*b('B4')-7.5*b('B3')+1))\",\n",
    "                      \"EVI2    = 2.5*((b('B8A')-b('B4'))/(b('B8A')+2.4*b('B4')+1))\",\n",
    "                      \"NLI     = ((b('B8A')**2)-b('B4'))/((b('B8A')**2)+b('B4'))\"]\n",
    "\n",
    "VI_list = [VI_definition.split()[0] for VI_definition in input_VI_definition]\n",
    "feature_bands = list(set(sentinel_0001 + sentinel_001 + sentinel_bands + VI_list) - set(sentinel_empty))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 356,
   "metadata": {},
   "outputs": [],
   "source": [
    "def format_sentinel_image(image):\n",
    "    image = ee.Image(image)\n",
    "    return image.select(sentinel_0001).multiply(0.0001) \\\n",
    "            .addBands(image.select(sentinel_001).multiply(0.001)) \\\n",
    "            .addBands(image.metadata(\"MEAN_SOLAR_ZENITH_ANGLE\").rename(\"SZA\").multiply(3.1415).divide(180).cos()) \\\n",
    "            .addBands(image.metadata(\"MEAN_SOLAR_AZIMUTH_ANGLE\").rename(\"SAA\").multiply(3.1415).divide(180).cos()) \\\n",
    "            .addBands(ee.Image(1).rename(\"ALT\")) \\\n",
    "            .addBands(image.select(sentinel_bands)) \\\n",
    "            .clip(filter_bounds) \\\n",
    "            .copyProperties(source=image)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 357,
   "metadata": {},
   "outputs": [],
   "source": [
    "def format_sentinel_collection(image_collection):\n",
    "    image_collection = ee.ImageCollection(image_collection)\n",
    "    \n",
    "    def format_sentinel_image(image):\n",
    "        image = ee.Image(image)\n",
    "        return image.select(sentinel_0001).multiply(0.0001) \\\n",
    "                .addBands(image.select(sentinel_001).multiply(0.001)) \\\n",
    "                .addBands(image.metadata(\"MEAN_SOLAR_ZENITH_ANGLE\").rename(\"SZA\").multiply(3.1415).divide(180).cos()) \\\n",
    "                .addBands(image.metadata(\"MEAN_SOLAR_AZIMUTH_ANGLE\").rename(\"SAA\").multiply(3.1415).divide(180).cos()) \\\n",
    "                .addBands(ee.Image(1).rename(\"ALT\")) \\\n",
    "                .addBands(image.select(sentinel_bands)) \\\n",
    "                .clip(filter_bounds) \\\n",
    "                .copyProperties(source=image)\n",
    "    \n",
    "    return image_collection.map(format_sentinel_image)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 358,
   "metadata": {},
   "outputs": [],
   "source": [
    "def add_VIs(image, VI_definition_list):\n",
    "    image = ee.Image(image)\n",
    "    VI_definition_list = ee.List(VI_definition_list)\n",
    "\n",
    "    # Generate an imageCollection from a list of expressions defining a set of Vegetation Indices using the bands available in the image\n",
    "    VIimage = ee.ImageCollection(VI_definition_list.map(lambda expr: image.expression(ee.String(expr)))).toBands().regexpRename(\"[0-9]+_\", \"\")\n",
    "\n",
    "    return image.addBands(VIimage)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 359,
   "metadata": {},
   "outputs": [],
   "source": [
    "def add_VIs_collection(image_collection, VI_definition_list, num_images: int = 0):\n",
    "    image_collection = ee.ImageCollection(image_collection)\n",
    "    VI_definition_list = ee.List(VI_definition_list)\n",
    "    \n",
    "    image_list = image_collection.toList(num_images)\n",
    "    collection = ee.List([])\n",
    "    \n",
    "    for i in range(num_images):\n",
    "        image = ee.Image(image_list.get(i))\n",
    "        VI_image = add_VIs(image, VI_definition_list)\n",
    "        collection = collection.add(VI_image)\n",
    "    \n",
    "    return collection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 360,
   "metadata": {},
   "outputs": [],
   "source": [
    "def centre_bands(image, bands, max_pixels: int = 1e8):\n",
    "    image = ee.Image(image)\n",
    "    \n",
    "    select_image = image.select(bands)\n",
    "    \n",
    "    mean_image = select_image.subtract(\n",
    "                    select_image.reduceRegion(reducer=ee.Reducer.mean(), geometry=select_image.geometry(), bestEffort=True, maxPixels=max_pixels)\n",
    "                    .toImage(bands)\n",
    "                    )\n",
    "    \n",
    "    return mean_image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 361,
   "metadata": {},
   "outputs": [],
   "source": [
    "def centre_bands_collection(image_collection, bands, max_pixels: int = 1e8):\n",
    "    image_collection = ee.ImageCollection(image_collection)\n",
    "    \n",
    "    def centre_bands_image(image):\n",
    "        image = ee.Image(image)\n",
    "\n",
    "        select_image = image.select(bands)\n",
    "\n",
    "        mean_image = select_image.subtract(\n",
    "                        select_image.reduceRegion(reducer=ee.Reducer.mean(), geometry=select_image.geometry(), bestEffort=True, maxPixels=max_pixels)\n",
    "                        .toImage(bands)\n",
    "                        )\n",
    "\n",
    "        return mean_image\n",
    "    \n",
    "    return image_collection.map(centre_bands_image)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 362,
   "metadata": {},
   "outputs": [],
   "source": [
    "def scale_bands(image, bands, max_pixels: int = 1e8):\n",
    "    image = ee.Image(image)\n",
    "    \n",
    "    mean_image = ee.Image(centre_bands(image, bands, max_pixels))\n",
    "    \n",
    "    standard_image = mean_image.divide(\n",
    "                        mean_image.reduceRegion(reducer=ee.Reducer.stdDev(), geometry=mean_image.geometry(), bestEffort=True, maxPixels=max_pixels)\n",
    "                        .toImage(bands)\n",
    "                        )\n",
    "    \n",
    "    return standard_image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 363,
   "metadata": {},
   "outputs": [],
   "source": [
    "def scale_bands_collection(image_collection, bands, max_pixels: int = 1e8):\n",
    "    image_collection = ee.ImageCollection(image_collection)\n",
    "    \n",
    "    def scale_bands(image):\n",
    "        image = ee.Image(image)\n",
    "\n",
    "        mean_image = ee.Image(centre_bands(image, bands, max_pixels))\n",
    "\n",
    "        standard_image = mean_image.divide(\n",
    "                            mean_image.reduceRegion(reducer=ee.Reducer.stdDev(), geometry=mean_image.geometry(), bestEffort=True, maxPixels=max_pixels)\n",
    "                            .toImage(bands)\n",
    "                            )\n",
    "\n",
    "        return standard_image\n",
    "    \n",
    "    return image_collection.map(scale_bands)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 364,
   "metadata": {},
   "outputs": [],
   "source": [
    "# The following function applies a layer in the neural network as parsed in the parse_layer function to an input image with the selected features, and response.\n",
    "# To fully apply the neural network we apply this function using the .iterate() method in Earth Engine on the input image in place of a loop\n",
    "def apply_layer_tansig(layer, inputs):\n",
    "    layer = ee.ImageCollection(layer)\n",
    "    inputs = ee.Image(inputs)\n",
    "    \n",
    "    activation = layer.getString(\"activation\")\n",
    "\n",
    "    def apply_node(node):\n",
    "        node = ee.Image(node)\n",
    "        bias = node.getNumber(\"bias\")\n",
    "        \n",
    "        # Multiply by the weights of the node and add the bias of the node\n",
    "        result = inputs.multiply(node).reduce(ee.Reducer.sum()).add(bias)\n",
    "        # For now we can only apply the softsign activation function if the activation property of the layer is not \"linear\"\n",
    "        return ee.Algorithms.If(activation.compareTo(\"linear\"), tansig(result), result)\n",
    "    \n",
    "    node_outputs = layer.map(apply_node).toBands()\n",
    "    \n",
    "    return node_outputs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 440,
   "metadata": {},
   "outputs": [],
   "source": [
    "def apply_nnet(image, features, neural_net, output_name, layer_function, max_pixels: int = 1e8, scale: bool = True):\n",
    "    image = ee.Image(image)\n",
    "    features = ee.List(features)\n",
    "    neural_net = ee.List(neural_net)\n",
    "    output_name = ee.String(output_name)\n",
    "    \n",
    "    if scale:\n",
    "        return image.addBands(neural_net.iterate(layer_function, scale_bands(image, features, max_pixels))).rename(output_name)\n",
    "    else:\n",
    "        return image.addBands(neural_net.iterate(layer_function, image)).rename(output_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 441,
   "metadata": {},
   "outputs": [],
   "source": [
    "def apply_nnet_collection(image_collection, feature_bands, neural_net, output_name, layer_function, max_pixels: int = 1e8, scale: bool = True):\n",
    "    image_collection = ee.ImageCollection(image_collection)\n",
    "    neural_net = ee.List(neural_net)\n",
    "    output_name = ee.String(output_name)\n",
    "    \n",
    "    if scale:\n",
    "        return image_collection.map(lambda image: image.addBands(ee.Image(neural_net.iterate(layer_function, scale_bands(image, feature_bands, max_pixels))).rename(output_name)))\n",
    "    else:\n",
    "        return image_collection.map(lambda image: image.addBands(ee.Image(neural_net.iterate(layer_function, image)).rename(output_name)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 367,
   "metadata": {},
   "outputs": [],
   "source": [
    "def ee_decile_buckets(inputs):\n",
    "    inputs = ee.FeatureCollection(inputs)\n",
    "    feature_bands = inputs.first().propertyNames().remove(\"system:index\")\n",
    "    \n",
    "    deciles = ee.List.sequence(0,100,10)\n",
    "    decile_names = deciles.map(lambda num: ee.Number(num).round().toInt().format(\"p%s\"))\n",
    "    decile_reducer = ee.Reducer.percentile(percentiles=ee.List.sequence(0, 100, 10), outputNames=decile_names, maxRaw=50000)\n",
    "\n",
    "    def get_decile_buckets(feature):\n",
    "        feature = ee.List(feature)\n",
    "\n",
    "        decile_boundaries = inputs.reduceColumns(reducer = decile_reducer, selectors = feature)\n",
    "        decile_buckets = ee.Dictionary.fromLists(ee.List.sequence(0,9).map(lambda num: ee.Number(num).round().toInt().format()), \n",
    "                                                decile_boundaries.values(decile_names.remove('p100')).zip(decile_boundaries.values(decile_names.remove('p0'))))\n",
    "        return decile_buckets\n",
    "\n",
    "    quantile_ranges_dict = ee.Dictionary.fromLists(feature_bands, ee.List(feature_bands.map(lambda feature: [feature])).map(get_decile_buckets))\n",
    "    \n",
    "    return quantile_ranges_dict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 368,
   "metadata": {},
   "outputs": [],
   "source": [
    "def ee_quality_codes(inputs, decile_buckets_dict):\n",
    "    inputs = ee.FeatureCollection(inputs)\n",
    "    decile_buckets_dict = ee.Dictionary(decile_buckets_dict)\n",
    "    \n",
    "    feature_bands = inputs.first().propertyNames().remove(\"system:index\")\n",
    "    \n",
    "    scale_codes = ee.Array(ee.List.sequence(feature_bands.size().subtract(1), 0, -1).map(lambda place: ee.Number(10).pow(place)))\n",
    "    \n",
    "    def code_feature(feature):\n",
    "        sample = ee.Feature(feature)\n",
    "        \n",
    "        def get_code_dict(key, value):\n",
    "            property_name = ee.String(key)\n",
    "            decile_buckets = ee.Dictionary(value)\n",
    "\n",
    "            sample_val = sample.getNumber(property_name)\n",
    "\n",
    "            def get_decile(decile_num):\n",
    "                decile_num = ee.Number(decile_num)\n",
    "                decile_bounds = ee.List(decile_buckets.get(decile_num.round().toInt().format()))\n",
    "\n",
    "                decile_min = decile_bounds.getNumber(0)\n",
    "                decile_max = decile_bounds.getNumber(1)\n",
    "\n",
    "                return sample_val.gt(decile_min).And(sample_val.lte(decile_max)).multiply(decile_num)\n",
    "\n",
    "            return ee.List.sequence(0,9).map(get_decile).reduce(ee.Reducer.sum())\n",
    "\n",
    "        return decile_buckets_dict.map(get_code_dict).toArray(keys=feature_bands).multiply(scale_codes).toList().reduce(ee.Reducer.sum())\n",
    "    \n",
    "    return inputs.toList(inputs.size()).map(code_feature)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 464,
   "metadata": {},
   "outputs": [],
   "source": [
    "def gen_quality_layer(image, band_decile_buckets, quality_codes, features, response, response_range, band_suffix = \"\"):\n",
    "    image = ee.Image(image)\n",
    "    band_decile_buckets = ee.Dictionary(band_decile_buckets)\n",
    "    quality_codes = ee.List(quality_codes)\n",
    "    features = ee.List(features)\n",
    "    response = ee.String(response)\n",
    "    response_range = ee.List(response_range)\n",
    "    \n",
    "    response_min = response_range.getNumber(0)\n",
    "    response_max = response_range.getNumber(1)\n",
    "    \n",
    "    scale_codes = ee.ImageCollection(ee.List.sequence(features.size().subtract(1), 0, -1).map(lambda place: ee.Image(ee.Number(10).pow(place)))) \\\n",
    "                    .toBands().rename(features)\n",
    "        \n",
    "    response_image = image.select(response)\n",
    "    above_min = response_image.gte(response_min).rename(\"above_min\")\n",
    "    below_max = response_image.lte(response_max).rename(\"below_max\")\n",
    "    pixel_in_range = above_min.And(below_max).rename(\"in_range\"+band_suffix)\n",
    "\n",
    "    def code_image(feature):\n",
    "        feature = ee.String(feature)\n",
    "\n",
    "        feature_quantiles = ee.Dictionary(band_decile_buckets.get(feature))\n",
    "        num_quantiles = feature_quantiles.size()\n",
    "\n",
    "        feature_image = image.select(feature)\n",
    "\n",
    "        feature_min = ee.List(feature_quantiles.get(\"0\")).getNumber(0)\n",
    "        below_min = feature_image.lte(feature_min).rename(\"below_min\")\n",
    "\n",
    "        feature_max = ee.List(feature_quantiles.get(num_quantiles.subtract(1).format())).getNumber(1)\n",
    "        above_max = feature_image.gt(feature_max).rename(\"above_max\")\n",
    "\n",
    "        out_of_range = above_max.Or(below_min).multiply(-1).rename(\"out_of_domain\")\n",
    "\n",
    "        def is_in_range(quantile):\n",
    "            quantile = ee.Number(quantile).toInt()\n",
    "            quantile_range = ee.List(feature_quantiles.get(quantile.format()))\n",
    "\n",
    "            quantile_range_min = quantile_range.getNumber(0)\n",
    "            quantile_range_max = quantile_range.getNumber(1)\n",
    "\n",
    "            return feature_image.gt(quantile_range_min).And(feature_image.lte(quantile_range_max)).multiply(quantile).rename(quantile.format())\n",
    "\n",
    "        return ee.ImageCollection(ee.List.sequence(0, num_quantiles.subtract(1)).map(is_in_range)).toBands() \\\n",
    "                .addBands(out_of_range).reduce(ee.Reducer.sum()).rename(feature)\n",
    "\n",
    "    pixel_quality_bands = ee.ImageCollection(features.map(code_image)).toBands() \\\n",
    "                            .rename(features.map(lambda feature: ee.String(feature).cat(\"_decile\"+band_suffix)))\n",
    "    pixel_in_domain = pixel_quality_bands.gt(-1).reduce(ee.Reducer.allNonZero()) \\\n",
    "                        .rename(\"in_domain\"+band_suffix)\n",
    "    pixel_quality_codes = pixel_quality_bands.multiply(scale_codes).reduce(ee.Reducer.sum()).multiply(pixel_in_domain.And(pixel_in_range).multiply(2).subtract(1)) \\\n",
    "                            .rename(\"quality_code\"+band_suffix)\n",
    "    pixel_quality = pixel_quality_codes.remap(quality_codes, ee.List.repeat(1, quality_codes.length())).mask() \\\n",
    "                    .rename(\"in_calibration_hypercubes\"+band_suffix)\n",
    "\n",
    "    quality_layer = pixel_in_domain.addBands(pixel_in_range).addBands(pixel_quality).addBands(pixel_quality_codes).addBands(pixel_quality_bands)\n",
    "    # This quality layer includes the deciles to which each band value belongs to from the calibration dataset at each pixel in the image. \n",
    "    # A value of 10 or 11 signifies that the band value at that pixel is out of the range of the values seen for the band in the calibration dataset. \n",
    "    # The in_calibration band is a 1 if the input combination of the band was found in the hypercubes from the calibration dataset and masked if it is not.\n",
    "    \n",
    "    return image.addBands(quality_layer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 465,
   "metadata": {},
   "outputs": [],
   "source": [
    "def gen_quality_layer_collection(image_collection, band_decile_buckets, quality_codes, features, response, response_range, band_suffix = \"\"):\n",
    "    image_collection = ee.ImageCollection(image_collection)\n",
    "    band_decile_buckets = ee.Dictionary(band_decile_buckets)\n",
    "    quality_codes = ee.List(quality_codes)\n",
    "    features = ee.List(features)\n",
    "    response = ee.String(response)\n",
    "    response_range = ee.List(response_range)\n",
    "    \n",
    "    response_min = response_range.getNumber(0)\n",
    "    response_max = response_range.getNumber(1)\n",
    "    \n",
    "    scale_codes = ee.ImageCollection(ee.List.sequence(features.size().subtract(1), 0, -1).map(lambda place: ee.Image(ee.Number(10).pow(place)))) \\\n",
    "                    .toBands().rename(features)\n",
    "        \n",
    "    def add_quality_layer(image):\n",
    "        image = ee.Image(image)\n",
    "        \n",
    "        response_image = image.select(response)\n",
    "        above_min = response_image.gte(response_min).rename(\"above_min\")\n",
    "        below_max = response_image.lte(response_max).rename(\"below_max\")\n",
    "        pixel_in_range = above_min.And(below_max).rename(\"in_range\"+band_suffix)\n",
    "\n",
    "        def code_image(feature):\n",
    "            feature = ee.String(feature)\n",
    "\n",
    "            feature_quantiles = ee.Dictionary(band_decile_buckets.get(feature))\n",
    "            num_quantiles = feature_quantiles.size()\n",
    "\n",
    "            feature_image = image.select(feature)\n",
    "\n",
    "            feature_min = ee.List(feature_quantiles.get(\"0\")).getNumber(0)\n",
    "            below_min = feature_image.lte(feature_min).rename(\"below_min\")\n",
    "\n",
    "            feature_max = ee.List(feature_quantiles.get(num_quantiles.subtract(1).format())).getNumber(1)\n",
    "            above_max = feature_image.gt(feature_max).rename(\"above_max\")\n",
    "\n",
    "            out_of_range = above_max.Or(below_min).multiply(-1).rename(\"out_of_domain\")\n",
    "\n",
    "            def is_in_range(quantile):\n",
    "                quantile = ee.Number(quantile).toInt()\n",
    "                quantile_range = ee.List(feature_quantiles.get(quantile.format()))\n",
    "\n",
    "                quantile_range_min = quantile_range.getNumber(0)\n",
    "                quantile_range_max = quantile_range.getNumber(1)\n",
    "\n",
    "                return feature_image.gt(quantile_range_min).And(feature_image.lte(quantile_range_max)).multiply(quantile).rename(quantile.format())\n",
    "\n",
    "            return ee.ImageCollection(ee.List.sequence(0, num_quantiles.subtract(1)).map(is_in_range)).toBands() \\\n",
    "                    .addBands(out_of_range).reduce(ee.Reducer.sum()).rename(feature)\n",
    "\n",
    "        pixel_quality_bands = ee.ImageCollection(features.map(code_image)).toBands() \\\n",
    "                                .rename(features.map(lambda feature: ee.String(feature).cat(\"_decile\"+band_suffix)))\n",
    "        pixel_in_domain = pixel_quality_bands.gt(-1).reduce(ee.Reducer.allNonZero()) \\\n",
    "                            .rename(\"in_domain\"+band_suffix)\n",
    "        pixel_quality_codes = pixel_quality_bands.multiply(scale_codes).reduce(ee.Reducer.sum()).multiply(pixel_in_domain.And(pixel_in_range).multiply(2).subtract(1)) \\\n",
    "                                .rename(\"quality_code\"+band_suffix)\n",
    "        pixel_quality = pixel_quality_codes.remap(quality_codes, ee.List.repeat(1, quality_codes.length())).mask() \\\n",
    "                        .rename(\"in_calibration_hypercubes\"+band_suffix)\n",
    "        \n",
    "        quality_layer = pixel_in_domain.addBands(pixel_in_range).addBands(pixel_quality).addBands(pixel_quality_codes).addBands(pixel_quality_bands)\n",
    "        # This quality layer includes the deciles to which each band value belongs to from the calibration dataset at each pixel in the image. \n",
    "        # A value of 10 or 11 signifies that the band value at that pixel is out of the range of the values seen for the band in the calibration dataset. \n",
    "        # The in_calibration band is a 1 if the input combination of the band was found in the hypercubes from the calibration dataset and masked if it is not.\n",
    "\n",
    "        return image.addBands(quality_layer)\n",
    "    \n",
    "    return image_collection.map(add_quality_layer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 371,
   "metadata": {},
   "outputs": [],
   "source": [
    "# The following function implements the LARs algorithm fully as described in ( et al. 2002)\n",
    "# It takes an image, a string which is the name of the band containing the response variable in the image, the number of non-zero\n",
    "# coefficients requested for the LARs algorithm to select the best features to predict the response in the image\n",
    "# Additionally the function requires the number of samples (pixels) from the image that the user wishes to process. \n",
    "# These inputs are necessary as Earth Engine provides a limited amount of RAM (2GB) and processing time on their virtual machines,\n",
    "# so the user may need to adjust how many pixels they wish to process in the image in case the function leads to a \n",
    "# \"User memory limit exceeded error\" or \"Computation timed out error\"\n",
    "def ee_LARS(sample_collection, features, response, num_nonzero_coefficients):\n",
    "    sample_collection = ee.FeatureCollection(sample_collection)\n",
    "    features = ee.List(features)\n",
    "    response = ee.String(response)\n",
    "    num_nonzero_coefficients = ee.Number(num_nonzero_coefficients)\n",
    "    \n",
    "    # Retrieving the list of features in the image by getting all of the band names and removing the response band name from the list\n",
    "    bandList = features.add(response)\n",
    "        \n",
    "    n = sample_collection.size()\n",
    "    m = features.length()\n",
    "    \n",
    "    # Using an aggregate array function over the feature collection and mapping the function over each feature in the band list\n",
    "    # to generate a dictionary of all of the samples we have retrieved\n",
    "    inputs = ee.Dictionary.fromLists(bandList, bandList.map(lambda feature: sample_collection.aggregate_array(feature)))\n",
    "    \n",
    "    # Although we may call our scale_image function on the input image, the reduceRegion function used to determine the mean\n",
    "    # and standard deviation of each band in the image over the entire region is not precise enough over a large image\n",
    "    # So we must recentre all of the bands in the image and now we can also normalize (L2 norm) each input feature as is required\n",
    "    # by the LARs algorithm\n",
    "    \n",
    "    # Also using an aggregate_mean function over the feature collection to get the mean of each band\n",
    "    input_means = ee.Dictionary.fromLists(bandList, bandList.map(lambda feature: sample_collection.aggregate_mean(feature)))\n",
    "    # Centering all of the bands in the image by mapping a function over the list of features and then a subtraction over the list of all of the samples for each band\n",
    "    def centre_inputs(key, value):\n",
    "        key_mean = input_means.getNumber(key)\n",
    "        return ee.List(value).map(lambda sample: ee.Number(sample).subtract(key_mean))\n",
    "    inputs = inputs.map(centre_inputs)\n",
    "    # Separating the response variable samples into its own vector\n",
    "    y = inputs.toArray([response]).reshape([-1,1])\n",
    "    \n",
    "    # Removing the response band from the feature collection by selecting only bands in the feature list\n",
    "    inputs = inputs.select(features)\n",
    "    # Generating a dictionary of all of the L2 norms of the input features using a custom mapped function\n",
    "    input_norms = inputs.map(lambda key, value: ee.Number(ee.List(value).map(lambda sample: ee.Number(sample).pow(2)).reduce(ee.Reducer.sum())).pow(0.5))\n",
    "    # Normalizing all of the features by mapping a function over the list of features and then mapping a division over the list of all of the samples of the feature\n",
    "    def norm_inputs(key, value):\n",
    "        key_norm = input_norms.getNumber(key)\n",
    "        return ee.List(value).map(lambda sample: ee.Number(sample).divide(key_norm))\n",
    "    inputs = inputs.map(norm_inputs)\n",
    "    # Generating the array of samples using the dictionary\n",
    "    X = inputs.toArray(features).transpose()\n",
    "    \n",
    "    \n",
    "    # Finding the first best predictor of the response to initialize the main LARs loop\n",
    "    initial_prediction = ee.Array(ee.List.repeat([0], n))\n",
    "    c = X.transpose().matrixMultiply(y.subtract(initial_prediction))\n",
    "    c_abs = c.abs()\n",
    "    C_maxLoc = c_abs.project([0]).argmax()\n",
    "    add_feature = C_maxLoc.getNumber(0)\n",
    "    A = ee.List([add_feature])\n",
    "    \n",
    "    # Creating a dicitionary of initial inputs to pass into the main LARs iterative loop\n",
    "    # The iterate function in Earth Engine process each iteration as a tree of iterations with no access to the any variables\n",
    "    # from previous iterations, only those that are passed to the next iteration, so we must pass both the current prediction and\n",
    "    # the active set of features (with non-zero coefficients), A\n",
    "    initial_inputs = ee.Dictionary({'prediction': initial_prediction, 'A': A})\n",
    "    \n",
    "    def LARs_regression(iteration, inputs):\n",
    "        inputs = ee.Dictionary(inputs)\n",
    "        \n",
    "        # Finding the active set of features, A (predictors with non-zero coefficients)\n",
    "        A = ee.List(inputs.get('A'))\n",
    "        # A_list is an array used to mask the full array of input samples and the correlation vector\n",
    "        A_list = ee.Array(ee.List.sequence(0, m.subtract(1)).map(lambda index: A.contains(index)).replaceAll(False, 0).replaceAll(True, 1)).reshape([-1,1])\n",
    "        \n",
    "        # The following matrix algebra determines the next most correlated variable, or the next best predictor considering the\n",
    "        # current features in the active set, A, as well as the magnitude to adjust the prediction vector to ensure all of the features\n",
    "        # in the active set are equally correlated to response vector\n",
    "        prediction = inputs.getArray('prediction')\n",
    "        c = X.transpose().matrixMultiply(y.subtract(prediction))\n",
    "        c_abs = c.abs()\n",
    "        C_max = c_abs.get(c_abs.argmax())\n",
    "        \n",
    "        s_A = c.divide(c_abs).mask(A_list)\n",
    "        X_A = X.mask(A_list.transpose())\n",
    "        G_Ai = X_A.transpose().matrixMultiply(X_A).matrixInverse()\n",
    "        G1 = G_Ai.matrixMultiply(s_A)\n",
    "        A_A = s_A.project([0]).dotProduct(G1.project([0])).pow(-0.5)\n",
    "        w_A = G1.multiply(A_A)\n",
    "        u_A = X_A.matrixMultiply(w_A)\n",
    "        a = X.transpose().matrixMultiply(u_A)\n",
    "        \n",
    "        a = a.project([0])\n",
    "        c = c.project([0])\n",
    "        def compute_gammaArray(index_j):\n",
    "            minus_j = C_max.subtract(c.get([index_j])).divide(A_A.subtract(a.get([index_j])))\n",
    "            plus_j = C_max.add(c.get([index_j])).divide(A_A.add(a.get([index_j])))\n",
    "            \n",
    "            return ee.List([minus_j, plus_j]).filter(ee.Filter.gte('item', 0)).reduce(ee.Reducer.min())\n",
    "        \n",
    "        A_c = ee.List.sequence(0, m.subtract(1)).removeAll(A)\n",
    "        gammaArray = A_c.map(compute_gammaArray)\n",
    "        gamma = gammaArray.reduce(ee.Reducer.min())\n",
    "        min_location = gammaArray.indexOf(gamma)\n",
    "        add_feature = A_c.getNumber(min_location)\n",
    "        \n",
    "        # Updating our active set of variables with the next best predictor from the non-active set and updating the prediction vector\n",
    "        A = A.add(add_feature)\n",
    "        prediction = prediction.add(u_A.multiply(gamma))\n",
    "        \n",
    "        return ee.Dictionary({'prediction': prediction, 'A': A})\n",
    "    \n",
    "    \n",
    "    # For the final iteration of LARs(if selecting all of the input variables) we require a different method to determine the magnitude\n",
    "    # to adjust the magnitude of the prediction vector, as the regular LARs iteration relies on the variables in the non-active set\n",
    "    # In the final iteration there will be no variables in the non-active set, so the method will not work\n",
    "    def LARs_final_iteration(iteration, inputs):\n",
    "        inputs = ee.Dictionary(inputs)\n",
    "        A = ee.List(inputs.get('A'))\n",
    "        \n",
    "        prediction = inputs.getArray('prediction')\n",
    "        c = X.transpose().matrixMultiply(y.subtract(prediction))\n",
    "        c_abs = c.abs()\n",
    "        C_max = c_abs.get(c_abs.argmax())        \n",
    "        \n",
    "        s_A = c.divide(c_abs)\n",
    "        G_Ai = X.transpose().matrixMultiply(X).matrixInverse()\n",
    "        G1 = G_Ai.matrixMultiply(s_A)\n",
    "        A_A = s_A.project([0]).dotProduct(G1.project([0])).pow(-0.5)\n",
    "        w_A = G1.multiply(A_A)\n",
    "        u_A = X.matrixMultiply(w_A)\n",
    "        \n",
    "        gamma = C_max.divide(A_A)\n",
    "        prediction = prediction.add(u_A.multiply(gamma))\n",
    "        \n",
    "        return ee.Dictionary({'prediction': prediction, 'A': A})\n",
    "    \n",
    "    # Actually carrying out the iterations by iterating over a placeholder list which is a sequence from 1 to the number of non-zero\n",
    "    # variables that the user wishes to select as predictors for the response\n",
    "    iterations = ee.List.sequence(1, m.subtract(1).min(num_nonzero_coefficients))\n",
    "    penultimate_outputs = iterations.iterate(LARs_regression, initial_inputs)\n",
    "    final_outputs = ee.Dictionary(ee.Algorithms.If(num_nonzero_coefficients.gte(m), LARs_final_iteration(m, penultimate_outputs), penultimate_outputs))\n",
    "    \n",
    "    final_prediction = final_outputs.getArray('prediction')\n",
    "    A = ee.List(final_outputs.get('A'))\n",
    "    feature_path = A.slice(0, num_nonzero_coefficients).map(lambda index: features.getString(index))\n",
    "    \n",
    "    # The code snippet below is able to extract the exact coefficients on all of the selected features, but is commented out\n",
    "    # as it adds computational complexity that takes up unnecessary memory on the Google Earth Engine virtual machine since we\n",
    "    # are only using LARs as a feature selection algorithm\n",
    "#     coefficients = X.matrixSolve(final_prediction).project([0]).toList().map(lambda num: ee.Algorithms.If(ee.Number(num).abs().lt(0.001), 0, num))\n",
    "#     print('Coefficients')\n",
    "#     coeff = ee.Dictionary.fromLists(features, coefficients).getInfo()\n",
    "#     ordered_coeff = OrderedDict()\n",
    "#     var_path = feature_path.cat(features.removeAll(feature_path)).getInfo()\n",
    "#     for key in var_path:\n",
    "#         ordered_coeff[key] = coeff[key]\n",
    "#     print(json.dumps(ordered_coeff, indent=1))\n",
    "    \n",
    "    return feature_path #.getInfo()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 482,
   "metadata": {},
   "outputs": [],
   "source": [
    "# The following function trims input data according to an algorithm in which the response band is partitioned into n equally sized\n",
    "# partitions, and in each of the n partitions, for the features selected by LARs, they are each trimmed individually down to only the\n",
    "# 5-95 percentile of the data. We are not doing any preprocessing with the data, so the raw data is exported from Earth Engine\n",
    "# The function takes an image, a list of strings with the selected feature bands in the image, the string that is the name of the response\n",
    "# band in this image, the number of samples/pixels the user wants to take from the image, and the number of parititions to trim within\n",
    "def trim_data(sample_collection, features, response, num_partitions):\n",
    "    sample_collection = ee.FeatureCollection(sample_collection)\n",
    "    features = ee.List(features)\n",
    "    response = ee.String(response)\n",
    "    num_partitions = ee.Number(num_partitions)\n",
    "    \n",
    "    # Generating the list of percentile bounds for the requested number of partitions, and the names of the value bounds for the\n",
    "    # dictionary that will be generated from the percentile reducer used later on\n",
    "    percentiles = ee.List.sequence(0, 100, ee.Number(100).divide(num_partitions))\n",
    "    percentile_names = percentiles.map(lambda num: ee.Number(num).round().toInt().format(\"p%s\"))\n",
    "    \n",
    "    # Finding the values at the percentile bounds using the percentile reducer over the feature collection\n",
    "    response_percentiles = sample_collection.reduceColumns(ee.Reducer.percentile(percentiles=percentiles, outputNames=percentile_names, maxRaw=sample_collection.size()), [response])\n",
    "    # Creating a list of percentile bounds for each partition\n",
    "    response_partitions = response_percentiles.values(percentile_names.remove('p100')).zip(response_percentiles.values(percentile_names.remove('p0')))\n",
    "    \n",
    "    # We'll use the following function mapped over the response_partitions list to partition the data by the requested number of partitions\n",
    "    def partition_data(partition_range):\n",
    "        partition_range = ee.List(partition_range)\n",
    "        return sample_collection.filter(ee.Filter.rangeContains(response, partition_range.getNumber(0), partition_range.getNumber(1)))\n",
    "    \n",
    "    partitioned_data = response_partitions.map(partition_data)\n",
    "    \n",
    "    # The following function now trims the data in each partition individually for each feature to its 5-95 percentile only\n",
    "    def trim_partitions(partition):\n",
    "        partition = ee.FeatureCollection(partition)\n",
    "        feature_trimming_bounds = features.map(lambda feature: ee.List([feature]).cat(partition.reduceColumns(ee.Reducer.percentile([5, 95]), [feature]).values(['p5','p95'])))\n",
    "        def trimmer(current_feature, collection):\n",
    "            current_feature = ee.List(current_feature)\n",
    "            collection = ee.FeatureCollection(collection)\n",
    "            return collection.filter(ee.Filter.rangeContains(current_feature.getString(0), current_feature.getNumber(1), current_feature.getNumber(2)))\n",
    "        return feature_trimming_bounds.iterate(trimmer, partition)\n",
    "    \n",
    "    # Retrieving the trimmed data partitions and flattening the paritions into a single trimmed feature collection\n",
    "    trimmed_partitions = partitioned_data.map(trim_partitions)\n",
    "    trimmed_data = ee.FeatureCollection(trimmed_partitions).flatten()\n",
    "    \n",
    "    return trimmed_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 446,
   "metadata": {},
   "outputs": [],
   "source": [
    "def apply_ALR_collection(image_collection, features, response, scale, max_pixels, num_lars_samples, num_lars_features, num_rf_samples, classifier):\n",
    "    image_collection = ee.ImageCollection(image_collection)\n",
    "    features = ee.List(features)\n",
    "    response = ee.String(response)\n",
    "    scale = ee.Number(scale)\n",
    "    max_pixels = ee.Number(max_pixels)\n",
    "    num_lars_samples = ee.Number(num_lars_samples)\n",
    "    num_lars_features = ee.Number(num_lars_features)\n",
    "    num_rf_samples = ee.Number(num_rf_samples)\n",
    "    classifier = ee.Classifier(classifier).setOutputMode(\"REGRESSION\")\n",
    "    \n",
    "    def apply_ALR_image(image):\n",
    "        image = ee.Image(image)\n",
    "        \n",
    "        lars_image = ee.Image(scale_bands(image, features, max_pixels)).addBands(ee.Image(centre_bands(image, [response], max_pixels))).clip(image.geometry())\n",
    "        lars_samples = lars_image.sample(region=lars_image.geometry(), scale=scale, numPixels=num_lars_samples)\n",
    "        \n",
    "        selected_features = ee_LARS(lars_samples, features, response, num_lars_features)\n",
    "        \n",
    "        rf_image = lars_image.select(selected_features).addBands(image.select(response)).clip(lars_image.geometry())\n",
    "        rf_samples = rf_image.sample(region=rf_image.geometry(), scale=scale, numPixels=num_rf_samples)\n",
    "        \n",
    "        rf_predictor = classifier.train(features = rf_samples, classProperty = response)\n",
    "        \n",
    "        rf_prediction = rf_image.classify(classifier = rf_predictor, outputName = output_var+\"_RF\")\n",
    "        rf_predicted_image = image.addBands(rf_prediction)\n",
    "        \n",
    "        rf_samples_deciles = ee_decile_buckets(rf_samples)\n",
    "        rf_quality_codes = ee_quality_codes(rf_samples, rf_samples_deciles)\n",
    "        rf_response_range = ee.List([rf_samples.aggregate_min(output_var), rf_samples.aggregate_max(output_var)])\n",
    "        rf_outputs = ee.Image(gen_quality_layer(rf_predicted_image, rf_samples_deciles, rf_quality_codes, selected_features, output_var+\"_RF\", rf_response_range, \"_RF\"))\n",
    "        \n",
    "        return rf_outputs\n",
    "    \n",
    "    return image_collection.map(apply_ALR_image)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 509,
   "metadata": {},
   "outputs": [],
   "source": [
    "def apply_ALR_with_trimming(image_collection, features, response, scale, max_pixels, num_lars_samples, num_lars_features, num_rf_samples, classifier):\n",
    "    image_collection = ee.ImageCollection(image_collection)\n",
    "    features = ee.List(features)\n",
    "    response = ee.String(response)\n",
    "    scale = ee.Number(scale)\n",
    "    max_pixels = ee.Number(max_pixels)\n",
    "    num_lars_samples = ee.Number(num_lars_samples)\n",
    "    num_lars_features = ee.Number(num_lars_features)\n",
    "    num_rf_samples = ee.Number(num_rf_samples)\n",
    "    classifier = ee.Classifier(classifier).setOutputMode(\"REGRESSION\")\n",
    "    \n",
    "    def apply_ALR_image(image):\n",
    "        image = ee.Image(image)\n",
    "        \n",
    "        lars_image = ee.Image(scale_bands(image, features, max_pixels)).addBands(ee.Image(centre_bands(image, [response], max_pixels))).clip(image.geometry())\n",
    "        lars_samples = lars_image.sample(region=lars_image.geometry(), scale=scale, numPixels=num_lars_samples)\n",
    "        \n",
    "        selected_features = ee_LARS(lars_samples, features, response, num_lars_features)\n",
    "        \n",
    "        rf_image = lars_image.select(selected_features).addBands(image.select(response)).clip(lars_image.geometry())\n",
    "        rf_sampled = rf_image.sample(region=rf_image.geometry(), scale=scale, numPixels=num_rf_samples)\n",
    "        \n",
    "        rf_samples = ee.FeatureCollection(trim_data(rf_sampled, selected_features, response, 10))\n",
    "        \n",
    "        rf_predictor = classifier.train(features = rf_samples, classProperty = response, inputProperties = selected_features)\n",
    "        \n",
    "        rf_prediction = rf_image.classify(classifier = rf_predictor, outputName = output_var+\"_RF\")\n",
    "        rf_predicted_image = image.addBands(rf_prediction)\n",
    "        \n",
    "        rf_samples_deciles = ee_decile_buckets(rf_samples)\n",
    "        rf_quality_codes = ee_quality_codes(rf_samples, rf_samples_deciles)\n",
    "        rf_response_range = ee.List([rf_samples.aggregate_min(output_var), rf_samples.aggregate_max(output_var)])\n",
    "        rf_outputs = ee.Image(gen_quality_layer(rf_predicted_image, rf_samples_deciles, rf_quality_codes, selected_features, output_var+\"_RF\", rf_response_range, \"_RF\"))\n",
    "        \n",
    "        return rf_outputs\n",
    "    \n",
    "    return image_collection.map(apply_ALR_image)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 466,
   "metadata": {},
   "outputs": [],
   "source": [
    "collection = \"COPERNICUS/S2_SR\"\n",
    "\n",
    "filter_bounds = ee.Geometry.Polygon(\n",
    "        [[[-98.16463919070031, 49.56174013736053],\n",
    "          [-97.74988035545425, 49.55800280566425],\n",
    "          [-97.74288559251306, 49.8299316015145],\n",
    "          [-98.15996345571678, 49.833705130736035]]])\n",
    "\n",
    "filter_start_date = ee.Date('2019-06-01')\n",
    "filter_end_date = ee.Date('2019-06-30')\n",
    "\n",
    "max_cloud_cover = 25"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 467,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_collection = ee.ImageCollection(collection).filterBounds(filter_bounds) \\\n",
    "                                                .filterDate(filter_start_date, filter_end_date) \\\n",
    "                                                .filterMetadata(\"WATER_PERCENTAGE\", \"less_than\", 100) \\\n",
    "                                                .filterMetadata(\"CLOUDY_PIXEL_PERCENTAGE\", \"less_than\", max_cloud_cover)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 468,
   "metadata": {},
   "outputs": [],
   "source": [
    "prepped_collection = ee.ImageCollection(format_sentinel_collection(image_collection = input_collection))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 469,
   "metadata": {},
   "outputs": [],
   "source": [
    "VI_collection = ee.ImageCollection(add_VIs_collection(image_collection   = prepped_collection, \n",
    "                                                      VI_definition_list = input_VI_definition, \n",
    "                                                      num_images         = num_input_images))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 470,
   "metadata": {},
   "outputs": [],
   "source": [
    "SL2P_predicted_collection = ee.ImageCollection(apply_nnet_collection(image_collection = VI_collection, \n",
    "                                                                     feature_bands    = features_SL2P, \n",
    "                                                                     neural_net       = SL2P, \n",
    "                                                                     output_name      = output_var, \n",
    "                                                                     layer_function   = apply_layer_tansig,\n",
    "                                                                     max_pixels       = 1e9,\n",
    "                                                                     scale            = True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 471,
   "metadata": {},
   "outputs": [],
   "source": [
    "SL2P_error_collection = ee.ImageCollection(apply_nnet_collection(image_collection = SL2P_predicted_collection, \n",
    "                                                                 feature_bands    = features_SL2P, \n",
    "                                                                 neural_net       = SL2P_error,\n",
    "                                                                 output_name      = output_var+\"_error\", \n",
    "                                                                 layer_function   = apply_layer_tansig,\n",
    "                                                                 max_pixels       = 1e9,\n",
    "                                                                 scale            = True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 472,
   "metadata": {},
   "outputs": [],
   "source": [
    "SL2P_output_collection = ee.ImageCollection(gen_quality_layer_collection(image_collection    = SL2P_error_collection, \n",
    "                                                                         band_decile_buckets = SL2P_input_deciles, \n",
    "                                                                         quality_codes       = SL2P_quality_codes, \n",
    "                                                                         features            = features_SL2P, \n",
    "                                                                         response            = output_var, \n",
    "                                                                         response_range      = SL2P_response_range, \n",
    "                                                                         band_suffix         = \"_SL2P\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 473,
   "metadata": {},
   "outputs": [],
   "source": [
    "random_forest = ee.Classifier.smileRandomForest(numberOfTrees = 100,\n",
    "                                                variablesPerSplit = 1, # Ensure that this is less than the number of features LARS is to select below\n",
    "                                                minLeafPopulation = 1,\n",
    "                                                bagFraction = 0.5).setOutputMode(\"REGRESSION\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 510,
   "metadata": {},
   "outputs": [],
   "source": [
    "ALR_output_collection = ee.ImageCollection(apply_ALR_collection(image_collection  = SL2P_output_collection, \n",
    "                                                                features          = feature_bands, \n",
    "                                                                response          = output_var, \n",
    "                                                                scale             = 20, \n",
    "                                                                max_pixels        = 1e9, \n",
    "                                                                num_lars_samples  = 10000,\n",
    "                                                                num_lars_features = 5, \n",
    "                                                                num_rf_samples    = 10000,\n",
    "                                                                classifier        = random_forest))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 511,
   "metadata": {},
   "outputs": [],
   "source": [
    "ALR_with_trimming_collection = ee.ImageCollection(apply_ALR_with_trimming(image_collection  = SL2P_output_collection, \n",
    "                                                                          features          = feature_bands, \n",
    "                                                                          response          = output_var, \n",
    "                                                                          scale             = 20, \n",
    "                                                                          max_pixels        = 1e9, \n",
    "                                                                          num_lars_samples  = 10000,\n",
    "                                                                          num_lars_features = 5, \n",
    "                                                                          num_rf_samples    = 10000,\n",
    "                                                                          classifier        = random_forest))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 475,
   "metadata": {},
   "outputs": [],
   "source": [
    "image_output_names = [name+\"_ALR_PREDICTION\" for name in input_collection.toList(input_collection.size()).map(lambda image: ee.Image(image).id()).getInfo()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 512,
   "metadata": {},
   "outputs": [],
   "source": [
    "ALR_export_tasks = export_collection_to_gee(collection   = ALR_output_collection, \n",
    "                                            num_images   = num_input_images, \n",
    "                                            image_names  = image_output_names, \n",
    "                                            asset_folder = \"users/hemitshah/ALR_output_collection\", \n",
    "                                            scale        = 20, \n",
    "                                            max_pixels   = 1e10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 514,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "20190603T173909_20190603T173911_T14UNV_ALR_PREDICTION: COMPLETED\n",
      "20190610T172909_20190610T173641_T14UNA_ALR_PREDICTION: COMPLETED\n",
      "20190610T172909_20190610T173641_T14UNV_ALR_PREDICTION: COMPLETED\n",
      "20190618T173911_20190618T174450_T14UNA_ALR_PREDICTION: COMPLETED\n",
      "20190625T172901_20190625T173920_T14UNV_ALR_PREDICTION: COMPLETED\n"
     ]
    }
   ],
   "source": [
    "check_ee_tasks(ALR_export_tasks)"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "name": "LAI_FAPAR_FCOVER_Prediction.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  },
  "pycharm": {
   "stem_cell": {
    "cell_type": "raw",
    "metadata": {
     "collapsed": false
    },
    "source": []
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}