gp_assignment.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "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": "3.6.7"
    },
    "latex_envs": {
      "bibliofile": "biblio.bib",
      "cite_by": "apalike",
      "current_citInitial": 1,
      "eqLabelWithNumbers": true,
      "eqNumInitial": 0
    },
    "colab": {
      "name": "gp_assignment.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "fAv83l5hLr1d"
      ]
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jomYta_gLrz0"
      },
      "source": [
        "# First things first\n",
        "Click **File -> Save a copy in Drive** and click **Open in new tab** in the pop-up window to save your progress in Google Drive."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mQRO0fzHLrz1"
      },
      "source": [
        "# Gaussian processes and Bayesian optimization"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "86OxrLRjLrz2"
      },
      "source": [
        "In this assignment you will learn how to use <a href=\"http://sheffieldml.github.io/GPy/\">GPy</a> and <a href=\"http://sheffieldml.github.io/GPyOpt/\">GPyOpt</a> libraries to deal with gaussian processes. These libraries provide quite simple and inuitive interfaces for training and inference, and we will try to get familiar with them in a few tasks."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OaSEPkjuLrz2"
      },
      "source": [
        "### Setup\n",
        "Load auxiliary files and then install and import the necessary libraries."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_o4qf1xhLrz3",
        "outputId": "3660cb7b-0c76-4b16-b342-dd2be7e792d8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 255
        }
      },
      "source": [
        "try:\n",
        "    import google.colab\n",
        "    IN_COLAB = True\n",
        "except:\n",
        "    IN_COLAB = False\n",
        "if IN_COLAB:\n",
        "    print(\"Downloading Colab files\")\n",
        "    ! shred -u setup_google_colab.py\n",
        "    ! wget https://raw.githubusercontent.com/hse-aml/bayesian-methods-for-ml/master/setup_google_colab.py -O setup_google_colab.py\n",
        "    import setup_google_colab\n",
        "    setup_google_colab.load_data_week6()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading Colab files\n",
            "shred: setup_google_colab.py: failed to open for writing: No such file or directory\n",
            "--2020-09-03 09:03:50--  https://raw.githubusercontent.com/hse-aml/bayesian-methods-for-ml/master/setup_google_colab.py\n",
            "Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 151.101.0.133, 151.101.64.133, 151.101.128.133, ...\n",
            "Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|151.101.0.133|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 1232 (1.2K) [text/plain]\n",
            "Saving to: ‘setup_google_colab.py’\n",
            "\n",
            "setup_google_colab. 100%[===================>]   1.20K  --.-KB/s    in 0s      \n",
            "\n",
            "2020-09-03 09:03:51 (65.6 MB/s) - ‘setup_google_colab.py’ saved [1232/1232]\n",
            "\n",
            "https://raw.githubusercontent.com/hse-aml/bayesian-methods-for-ml/master/week6/w6_grader.py w6_grader.py\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oMbwv_YgLrz9",
        "outputId": "b3d92aca-48db-45af-8405-9294b4fe24c0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 476
        }
      },
      "source": [
        "! pip install GPy gpyopt xgboost"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting GPy\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/67/95/976598f98adbfa918a480cb2d643f93fb555ca5b6c5614f76b69678114c1/GPy-1.9.9.tar.gz (995kB)\n",
            "\u001b[K     |████████████████████████████████| 1.0MB 1.8MB/s \n",
            "\u001b[?25hCollecting gpyopt\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/52/be/669d505416d7e465b2aef7df3b58d590f56468c4f7dc50c91fe91b8a78d9/GPyOpt-1.2.6.tar.gz (56kB)\n",
            "\u001b[K     |████████████████████████████████| 61kB 5.5MB/s \n",
            "\u001b[?25hRequirement already satisfied: xgboost in /usr/local/lib/python3.6/dist-packages (0.90)\n",
            "Requirement already satisfied: numpy>=1.7 in /usr/local/lib/python3.6/dist-packages (from GPy) (1.18.5)\n",
            "Requirement already satisfied: scipy>=0.16 in /usr/local/lib/python3.6/dist-packages (from GPy) (1.4.1)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from GPy) (1.15.0)\n",
            "Collecting paramz>=0.9.0\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/d8/37/4abbeb78d30f20d3402887f46e6e9f3ef32034a9dea65d243654c82c8553/paramz-0.9.5.tar.gz (71kB)\n",
            "\u001b[K     |████████████████████████████████| 71kB 5.3MB/s \n",
            "\u001b[?25hRequirement already satisfied: decorator>=4.0.10 in /usr/local/lib/python3.6/dist-packages (from paramz>=0.9.0->GPy) (4.4.2)\n",
            "Building wheels for collected packages: GPy, gpyopt, paramz\n",
            "  Building wheel for GPy (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for GPy: filename=GPy-1.9.9-cp36-cp36m-linux_x86_64.whl size=2633901 sha256=1316efa12efd9a6b3542e619ccf3930d687771f6d8f9cc739a9ce952e7a1a100\n",
            "  Stored in directory: /root/.cache/pip/wheels/5d/36/66/2b58860c84c9f2b51615da66bfd6feeddbc4e04d887ff96dfa\n",
            "  Building wheel for gpyopt (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for gpyopt: filename=GPyOpt-1.2.6-cp36-none-any.whl size=83622 sha256=755d1c6310a6d7a621251ce93306d2dfe608cbbd474c1c86effdd6d1bbeb7b9c\n",
            "  Stored in directory: /root/.cache/pip/wheels/b2/00/69/cfa967a125cf25e66f644be6193ad6f0edf231147879ad714f\n",
            "  Building wheel for paramz (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for paramz: filename=paramz-0.9.5-cp36-none-any.whl size=102551 sha256=dea379a8714c0d94923e1a913e901e6e8061f18c43c7c08499c5ceec1a2fbd78\n",
            "  Stored in directory: /root/.cache/pip/wheels/c8/4a/0e/6e0dc85541825f991c431619e25b870d4b812c911214690cf8\n",
            "Successfully built GPy gpyopt paramz\n",
            "Installing collected packages: paramz, GPy, gpyopt\n",
            "Successfully installed GPy-1.9.9 gpyopt-1.2.6 paramz-0.9.5\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "V1-xNQbcLr0B"
      },
      "source": [
        "import numpy as np\n",
        "import GPy\n",
        "import GPyOpt\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn.svm import SVR\n",
        "import sklearn.datasets\n",
        "from xgboost import XGBRegressor\n",
        "from sklearn.model_selection import cross_val_score\n",
        "import time\n",
        "from w6_grader import GPGrader\n",
        "%matplotlib inline"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "s1RKnIW_Lr0F"
      },
      "source": [
        "### Grading\n",
        "We will create a grader instace below and use it to collect your answers. Note that these outputs will be stored locally inside grader and will be uploaded to platform only after running submiting function in the last part of this assignment. If you want to make partial submission, you can run that cell any time you want."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fqqfps3FLr0F"
      },
      "source": [
        "grader = GPGrader()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0Omi6n9qLr0J"
      },
      "source": [
        "## Gaussian processes: GPy (<a href=\"http://pythonhosted.org/GPy/\">documentation</a>)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "collapsed": true,
        "id": "5XRSqkm-Lr0K"
      },
      "source": [
        "We will start with a simple regression problem, for which we will try to fit a Gaussian Process with RBF kernel."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-E9fIaN7Lr0L"
      },
      "source": [
        "def generate_points(n=25, noise_variance=0.0036):\n",
        "    np.random.seed(777)\n",
        "    X = np.random.uniform(-3., 3., (n, 1))\n",
        "    y = np.sin(X) + np.random.randn(n, 1) * noise_variance**0.5\n",
        "    return X, y\n",
        "    \n",
        "def generate_noise(n=25, noise_variance=0.0036):\n",
        "    np.random.seed(777)\n",
        "    X = np.random.uniform(-3., 3., (n, 1))\n",
        "    y = np.random.randn(n, 1) * noise_variance**0.5\n",
        "    return X, y"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "L0oT3wdiLr0P",
        "outputId": "fd9120ab-9bf9-42ca-d5f8-9c5096232861",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "source": [
        "# Create data points\n",
        "X, y = generate_points()\n",
        "plt.plot(X, y, '.')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ftQpKllSLr0V"
      },
      "source": [
        "To fit a Gaussian Process, you will need to define a kernel. For Gaussian (GBF) kernel you can use `GPy.kern.RBF` function."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iaDsV9QkLr0W"
      },
      "source": [
        "<b> Task 1.1: </b> Create RBF kernel with variance 1.5 and length-scale parameter 2 for 1D samples and compute value of the kernel between points `X[5]` and `X[9]`. Submit a single number. \n",
        "<br><b>Hint:</b> use `.K` property of kernel object."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "i1AcAncFLr0W",
        "outputId": "bbc769a5-b228-4133-c169-7d590cfada78",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "kernel = GPy.kern.RBF(input_dim=1,variance=1.5,lengthscale=2.) ### YOUR CODE HERE\n",
        "kernel_59 = kernel.K(X[5].reshape(-1,1),X[9].reshape(-1,1))[0] ### YOUR CODE HERE\n",
        "grader.submit_GPy_1(kernel_59)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Current answer for task 1.1 is: 1.0461813545396959\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lBNAh-eLLr0c"
      },
      "source": [
        "<b> Task 1.2: </b> Fit GP into generated data. Use kernel from previous task. Submit predicted mean and vairance at position $x=1$.\n",
        "<br><b>Hint:</b> use `GPy.models.GPRegression` class."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KN0HBIVrajRy"
      },
      "source": [
        "model = GPy.models.GPRegression(X,y,kernel=kernel)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dYFc7vrpakdh",
        "outputId": "f3220466-0f81-4fa4-ed6f-f027694ee486",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        }
      },
      "source": [
        "print(model['.*var'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "  \u001b[1mindex\u001b[0;0m  |             GP_regression.rbf.variance  |  constraints  |  priors\n",
            "  \u001b[1m[0]  \u001b[0;0m  |                             1.50000000  |      +ve      |        \n",
            "  \u001b[1m-----\u001b[0;0m  |  GP_regression.Gaussian_noise.variance  |  -----------  |  ------\n",
            "  \u001b[1m[0]  \u001b[0;0m  |                             1.00000000  |      +ve      |        \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YAFpYWPRcVJv",
        "outputId": "0ae27aa7-0e79-4eba-bbd7-c5daf1641215",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "model.predict(X[1].reshape(-1,1))[0]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[-0.690363]])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 92
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mxAqv6KRLr0d",
        "outputId": "a0ff615c-49b8-4add-c7f8-7c2e39748aac",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "model = GPy.models.GPRegression(X,y,kernel=kernel)\n",
        " ### YOUR CODE HERE\n",
        "preds=model.predict(np.array([[1]]))\n",
        "mean =  preds[0] ### YOUR CODE HERE\n",
        "variance = preds[1]### YOUR CODE HERE\n",
        "grader.submit_GPy_2(mean, variance)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Current answer for task 1.2 (mean) is: 0.6646774926102937\n",
            "Current answer for task 1.2 (variance) is: 1.1001478223790582\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XPSe9Nk7Lr0h",
        "outputId": "a1143044-42c2-449d-9de6-287adc291159",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        }
      },
      "source": [
        "model.plot()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zagLn1BdLr0l"
      },
      "source": [
        "We see that the model didn't fit the data quite well. Let's try to fit kernel and noise parameters automatically as discussed in the lecture! You can see the current parameters below:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "22D0ahSULr0m",
        "outputId": "6129f8c7-dbf8-4ecc-834a-f044ac800418",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 221
        }
      },
      "source": [
        "model"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<style type=\"text/css\">\n",
              ".pd{\n",
              "    font-family: \"Courier New\", Courier, monospace !important;\n",
              "    width: 100%;\n",
              "    padding: 3px;\n",
              "}\n",
              "</style>\n",
              "\n",
              "<p class=pd>\n",
              "<b>Model</b>: GP regression<br>\n",
              "<b>Objective</b>: 27.86687636693494<br>\n",
              "<b>Number of Parameters</b>: 3<br>\n",
              "<b>Number of Optimization Parameters</b>: 3<br>\n",
              "<b>Updates</b>: True<br>\n",
              "</p>\n",
              "<style type=\"text/css\">\n",
              ".tg  {font-family:\"Courier New\", Courier, monospace !important;padding:2px 3px;word-break:normal;border-collapse:collapse;border-spacing:0;border-color:#DCDCDC;margin:0px auto;width:100%;}\n",
              ".tg td{font-family:\"Courier New\", Courier, monospace !important;font-weight:bold;color:#444;background-color:#F7FDFA;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;border-color:#DCDCDC;}\n",
              ".tg th{font-family:\"Courier New\", Courier, monospace !important;font-weight:normal;color:#fff;background-color:#26ADE4;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;border-color:#DCDCDC;}\n",
              ".tg .tg-left{font-family:\"Courier New\", Courier, monospace !important;font-weight:normal;text-align:left;}\n",
              ".tg .tg-center{font-family:\"Courier New\", Courier, monospace !important;font-weight:normal;text-align:center;}\n",
              ".tg .tg-right{font-family:\"Courier New\", Courier, monospace !important;font-weight:normal;text-align:right;}\n",
              "</style>\n",
              "<table class=\"tg\"><tr><th><b>  GP_regression.         </b></th><th><b>value</b></th><th><b>constraints</b></th><th><b>priors</b></th></tr>\n",
              "<tr><td class=tg-left>  rbf.variance           </td><td class=tg-right>  1.5</td><td class=tg-center>    +ve    </td><td class=tg-center>      </td></tr>\n",
              "<tr><td class=tg-left>  rbf.lengthscale        </td><td class=tg-right>  2.0</td><td class=tg-center>    +ve    </td><td class=tg-center>      </td></tr>\n",
              "<tr><td class=tg-left>  Gaussian_noise.variance</td><td class=tg-right>  1.0</td><td class=tg-center>    +ve    </td><td class=tg-center>      </td></tr>\n",
              "</table>"
            ],
            "text/plain": [
              "<GPy.models.gp_regression.GPRegression at 0x7f6d717b0dd8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pj9UkrzBLr0q"
      },
      "source": [
        "<b> Task 1.3: </b> Optimize length-scale, variance and noise component of the model and submit optimal length-scale value of the kernel. \n",
        "<br><b>Hint:</b> Use `.optimize()` function of the model and `.lengthscale` property of the kernel."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oZ7qUxDlcyOu",
        "outputId": "0e6504c6-3bda-4043-ae13-7d2ab208444d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "model.optimize()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<paramz.optimization.optimization.opt_lbfgsb at 0x7f6d9671a7f0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u6Y0iDyGc40A",
        "outputId": "e3adc690-d799-4593-af78-88e74c5e22da",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 221
        }
      },
      "source": [
        "model"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<style type=\"text/css\">\n",
              ".pd{\n",
              "    font-family: \"Courier New\", Courier, monospace !important;\n",
              "    width: 100%;\n",
              "    padding: 3px;\n",
              "}\n",
              "</style>\n",
              "\n",
              "<p class=pd>\n",
              "<b>Model</b>: GP regression<br>\n",
              "<b>Objective</b>: -18.35176775416764<br>\n",
              "<b>Number of Parameters</b>: 3<br>\n",
              "<b>Number of Optimization Parameters</b>: 3<br>\n",
              "<b>Updates</b>: True<br>\n",
              "</p>\n",
              "<style type=\"text/css\">\n",
              ".tg  {font-family:\"Courier New\", Courier, monospace !important;padding:2px 3px;word-break:normal;border-collapse:collapse;border-spacing:0;border-color:#DCDCDC;margin:0px auto;width:100%;}\n",
              ".tg td{font-family:\"Courier New\", Courier, monospace !important;font-weight:bold;color:#444;background-color:#F7FDFA;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;border-color:#DCDCDC;}\n",
              ".tg th{font-family:\"Courier New\", Courier, monospace !important;font-weight:normal;color:#fff;background-color:#26ADE4;border-style:solid;border-width:1px;overflow:hidden;word-break:normal;border-color:#DCDCDC;}\n",
              ".tg .tg-left{font-family:\"Courier New\", Courier, monospace !important;font-weight:normal;text-align:left;}\n",
              ".tg .tg-center{font-family:\"Courier New\", Courier, monospace !important;font-weight:normal;text-align:center;}\n",
              ".tg .tg-right{font-family:\"Courier New\", Courier, monospace !important;font-weight:normal;text-align:right;}\n",
              "</style>\n",
              "<table class=\"tg\"><tr><th><b>  GP_regression.         </b></th><th><b>                value</b></th><th><b>constraints</b></th><th><b>priors</b></th></tr>\n",
              "<tr><td class=tg-left>  rbf.variance           </td><td class=tg-right>   0.7099385192642236</td><td class=tg-center>    +ve    </td><td class=tg-center>      </td></tr>\n",
              "<tr><td class=tg-left>  rbf.lengthscale        </td><td class=tg-right>   1.6252681650349912</td><td class=tg-center>    +ve    </td><td class=tg-center>      </td></tr>\n",
              "<tr><td class=tg-left>  Gaussian_noise.variance</td><td class=tg-right>0.0038978708233022826</td><td class=tg-center>    +ve    </td><td class=tg-center>      </td></tr>\n",
              "</table>"
            ],
            "text/plain": [
              "<GPy.models.gp_regression.GPRegression at 0x7f6d717b0dd8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 30
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vrHd6ZHXLr0q",
        "outputId": "e4d69fcb-a5f2-46b3-bf9e-5226d79429e6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "### YOUR CODE HERE\n",
        "lengthscale=model.kern.lengthscale\n",
        "grader.submit_GPy_3(lengthscale)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Current answer for task 1.3 is: 1.6252681650349912\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cGaxb-moLr0v",
        "outputId": "6c6fcabf-b225-460c-c379-a35f0f22b2eb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        }
      },
      "source": [
        "model.plot()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0HHx4BMGLr0y"
      },
      "source": [
        "As you see, the process generates outputs just right. Let's see if GP can figure out itself when we try to fit it into noise or signal."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xVPDSbpzLr0y"
      },
      "source": [
        "<b> Task 1.4: </b> Generate two datasets: sinusoid wihout noise and samples from gaussian noise. Optimize kernel parameters and submit optimal values of noise component.\n",
        "<br><b>Note:</b> generate data only using ```generate_points(n, noise_variance)``` and ```generate_noise(n, noise_variance)``` function!"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sqRYrAbSLr0z"
      },
      "source": [
        "X, y = generate_noise(noise_variance=10)\n",
        "k=GPy.kern.RBF(input_dim=1,)\n",
        "mod=GPy.models.GPRegression(X,y,k)\n",
        "mod.optimize()\n",
        "noise=mod.Gaussian_noise[0]\n",
        "### YOUR CODE HERE"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jxNkAgGCnN7U",
        "outputId": "a0af8664-a7cd-41d3-a889-5bd7717ade49",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "mod.Gaussian_noise[0]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "10.143215047293623"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 38
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nePxIA8iLr02"
      },
      "source": [
        "X, y = generate_points(noise_variance=0)\n",
        "k_n=GPy.kern.RBF(input_dim=1,)\n",
        "mod_n=GPy.models.GPRegression(X,y,k_n)\n",
        "mod_n.optimize()\n",
        "just_signal=mod_n.Gaussian_noise.variance\n",
        "### YOUR CODE HERE"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BfpMoREWLr0-",
        "outputId": "c3ae1acb-1c12-43b2-eb96-3fea3eac5b51",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "grader.submit_GPy_4(noise, just_signal)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Current answer for task 1.4 (noise) is: 10.143215047293623\n",
            "Current answer for task 1.4 (just signal) is: 3.686656201867028e-14\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ORekwW3ZLr1D"
      },
      "source": [
        "## Sparse GP\n",
        "Now let's consider the speed of GP. We will generate a dataset of 3000 points and measure the time that is consumed for prediction of mean and variance for each point. We will then try to use inducing inputs and find the optimal number of points according to quality-time tradeoff.\n",
        "\n",
        "For the sparse model with inducing points, you should use ```GPy.models.SparseGPRegression``` class. You can set the number of inducing inputs with parameter ```num_inducing``` and optimize their positions and values with ```.optimize()``` call."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZkISyG0VLr1E"
      },
      "source": [
        "<b>Task 1.5</b>: Create a dataset of 1000 points and fit GPRegression. Measure time for predicting mean and variance at position $x=1$. Then fit `SparseGPRegression` with 10 inducing inputs and repeat the experiment. Report speedup as a ratio between consumed time without and with inducing inputs."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6_JDMzdQLr1G"
      },
      "source": [
        "X, y = generate_points(1000)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6g5OaGlVLr1K"
      },
      "source": [
        "\n",
        "start = time.time()\n",
        "### YOUR CODE HERE\n",
        "ker=GPy.kern.RBF(input_dim=1)\n",
        "mod=GPy.models.GPRegression(X,y,ker)\n",
        "mod.optimize()\n",
        "mean,var=model.predict(X[1].reshape(-1,1))\n",
        "time_gp = time.time()-start"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VC2mj9bPLr1Q"
      },
      "source": [
        "start = time.time()\n",
        "### YOUR CODE HERE\n",
        "ker=GPy.kern.RBF(input_dim=1)\n",
        "mode=GPy.models.SparseGPRegression(X,y,kernel=ker,num_inducing=10)\n",
        "mode.optimize()\n",
        "mean_1,var_1=model.predict(X[1].reshape(-1,1))\n",
        "time_sgp = time.time()-start"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WxGxBPYJLr1T",
        "outputId": "a6442bb0-fd1a-457f-ef9c-d66bed9a3a06",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        }
      },
      "source": [
        "mod.plot()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yu9aE4ZRo4Zr",
        "outputId": "5a8afa99-cc51-4be3-d5a0-da6d7edbc804",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "print(time_gp,time_sgp)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "6.671249628067017 1.3362116813659668\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MtkY1cLLLr1a",
        "outputId": "d33f2406-3883-49fa-c0f3-08d8a1673118",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "grader.submit_GPy_5(time_gp / time_sgp)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Current answer for task 1.5 is: 4.992659262825191\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fAv83l5hLr1d"
      },
      "source": [
        "## Bayesian optimization: GPyOpt (<a href=\"http://pythonhosted.org/GPyOpt/\">documentation</a>, <a href=\"http://nbviewer.jupyter.org/github/SheffieldML/GPyOpt/blob/master/manual/index.ipynb\">tutorials</a>)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wH8NqfSgLr1e"
      },
      "source": [
        "In this part of the assignment, we will try to find optimal hyperparameters to XGBoost model! We will use data from a small competition to speed things up, but keep in mind that the approach works even for large datasets.\n",
        "\n",
        "We will use diabetes dataset provided in sklearn package."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fJPrIj7wLr1f"
      },
      "source": [
        "dataset = sklearn.datasets.load_diabetes()\n",
        "X = dataset['data']\n",
        "y = dataset['target']"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zQyVkmt5Lr1j"
      },
      "source": [
        "We will use cross-validation score to estimate accuracy and our goal will be to tune: ```max_depth```, ```learning_rate```, ```n_estimators``` parameters. The baseline MSE with default XGBoost parameters is $0.2$. Let's see if we can do better. First, we have to define optimization function and domains."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6sjEdwv-Lr1k"
      },
      "source": [
        "# Score. Optimizer will try to find minimum, so we will add a \"-\" sign.\n",
        "def f(parameters):\n",
        "    parameters = parameters[0]\n",
        "    score = -cross_val_score(\n",
        "        XGBRegressor(learning_rate=parameters[0],\n",
        "                     max_depth=int(parameters[2]),\n",
        "                     n_estimators=int(parameters[3]),\n",
        "                     gamma=int(parameters[1]),\n",
        "                     min_child_weight = parameters[4]), \n",
        "        X, y, scoring='neg_mean_squared_error'\n",
        "    ).mean()\n",
        "    score = np.array(score)\n",
        "    return score"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "196-YC1JLr1n",
        "outputId": "23682250-e4ce-486f-b9f2-b0b7aa9e8925",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "baseline = -cross_val_score(\n",
        "    XGBRegressor(), X, y, scoring='neg_mean_squared_error'\n",
        ").mean()\n",
        "baseline"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[11:38:54] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:54] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:54] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:54] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:54] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "3437.074407214106"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 96
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RoyJgS1pLr1r"
      },
      "source": [
        "# Bounds (NOTE: define continuous variables first, then discrete!)\n",
        "bounds = [\n",
        "    {'name': 'learning_rate',\n",
        "     'type': 'continuous',\n",
        "     'domain': (0, 1)},\n",
        "\n",
        "    {'name': 'gamma',\n",
        "     'type': 'continuous',\n",
        "     'domain': (0, 5)},\n",
        "\n",
        "    {'name': 'max_depth',\n",
        "     'type': 'discrete',\n",
        "     'domain': (1, 50)},\n",
        "\n",
        "    {'name': 'n_estimators',\n",
        "     'type': 'discrete',\n",
        "     'domain': (1, 300)},\n",
        "\n",
        "    {'name': 'min_child_weight',\n",
        "     'type': 'discrete',\n",
        "     'domain': (1, 10)}\n",
        "]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "scrolled": false,
        "id": "HbM2jYs0Lr1v",
        "outputId": "bcabeb31-9baf-458f-9d88-4c256e54a53b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 442
        }
      },
      "source": [
        "np.random.seed(777)\n",
        "optimizer = GPyOpt.methods.BayesianOptimization(\n",
        "    f=f, domain=bounds,\n",
        "    acquisition_type ='MPI',\n",
        "    acquisition_par = 0.1,\n",
        "    exact_eval=True\n",
        ")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[11:38:54] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:55] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:55] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:55] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:55] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:56] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:56] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:56] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:57] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:57] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:57] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:58] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:58] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:38:59] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:00] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:00] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:00] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:00] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:00] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:00] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:00] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:00] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:00] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:00] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:00] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "m6IZHeUHLr1x",
        "outputId": "0e3e06d6-ec42-4d61-ce0b-8bf426ad6960",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "max_iter = 50\n",
        "max_time = 60\n",
        "optimizer.run_optimization(max_iter, max_time)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[11:39:01] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:01] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:01] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:01] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:01] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:01] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:01] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:01] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:01] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:01] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:02] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:02] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:02] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:02] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:02] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:03] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:03] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:03] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:03] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:03] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:04] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:04] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:04] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:04] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:04] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:04] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:04] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:04] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:04] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:05] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:05] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:05] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:05] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:05] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:05] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:06] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:06] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:06] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:06] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:06] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:06] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:07] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:07] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:07] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:07] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:07] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:07] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:07] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:07] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:07] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:08] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:08] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:08] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:08] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:08] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:09] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:09] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:09] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:09] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:09] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:10] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:10] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:10] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:10] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:10] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:11] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:11] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:11] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:11] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:11] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:11] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:11] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:11] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:12] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:12] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:13] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:13] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:13] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:13] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:13] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:13] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:14] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:14] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:14] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:14] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:14] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:14] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:14] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:14] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:15] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:15] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:15] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:15] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:16] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:16] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:16] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:16] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:16] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:16] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:16] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:17] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:17] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:17] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:17] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:17] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:18] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:18] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:18] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:18] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:18] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:19] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:19] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:19] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:19] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:19] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:20] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:20] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:20] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:20] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:20] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:21] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:21] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:21] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:21] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:21] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:22] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:22] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:22] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:22] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:22] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:23] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:23] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:23] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:24] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:24] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:25] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:25] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:25] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:25] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:25] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:26] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:26] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:26] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:26] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:26] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:27] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:27] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:27] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:27] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:27] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:28] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:28] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:28] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:28] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:28] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:29] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:29] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:29] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:29] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:29] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:30] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:30] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:30] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:30] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:31] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:31] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:31] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:31] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:31] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:31] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:33] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:33] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:33] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:33] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:33] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:34] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:34] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:34] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:34] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:34] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:35] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:35] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:35] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:35] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:35] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:37] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:37] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:37] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:37] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:37] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:38] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:38] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:38] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:38] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:39] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:39] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:39] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:39] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:39] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:39] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:40] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:40] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:41] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:41] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:41] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:42] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:42] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:42] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:42] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:42] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:43] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:43] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:43] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:43] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:43] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:44] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:44] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:44] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:44] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:44] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:45] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:45] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:45] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:45] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:45] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:46] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:46] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:46] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:46] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:46] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:48] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:48] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:48] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:48] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:48] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:49] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:49] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:49] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:49] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:49] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:50] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:50] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:50] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:50] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:50] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:51] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:51] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:51] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:51] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n",
            "[11:39:51] WARNING: /workspace/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iWVqyrSWLr10",
        "outputId": "67c0dd50-054a-4952-82a5-54bee8ef73c8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 350
        }
      },
      "source": [
        "optimizer.plot_convergence()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x360 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LTKK2nZRLr15"
      },
      "source": [
        "Best values of parameters:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lGPtOBjuLr15",
        "outputId": "0519deb4-fb88-449d-d363-029c588e585e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "optimizer.X[np.argmin(optimizer.Y)]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([4.76325090e-02, 1.19202751e+00, 1.00000000e+00, 3.00000000e+02,\n",
              "       1.00000000e+00])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 101
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hzJTJ-IJLr18",
        "outputId": "2aa9515f-6f96-4b86-e187-33a6ffbeb2e3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "print('MSE:', np.min(optimizer.Y),\n",
        "      'Gain:', baseline/np.min(optimizer.Y)*100)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "MSE: 3189.185708011576 Gain: 107.77278973061391\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ePwYSkpeLr1-"
      },
      "source": [
        "We were able to get 9% boost without tuning parameters by hand! Let's see if you can do the same. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ef6sGsU4Lr1_"
      },
      "source": [
        "<b>Task 2.1:</b> Tune SVR model. Find optimal values for three parameters: `C`, `epsilon` and `gamma`. Use range (1e-5, 1000) for `C`, (1e-5, 10) for `epsilon` and `gamma`. Use MPI as an acquisition function with weight 0.1. Submit the optimal value of epsilon that was found by a model."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KM2l0ddIy3hx",
        "outputId": "a275001b-7d2b-4f9b-fc13-c2393ac602b8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "baseline = -cross_val_score(\n",
        "    SVR(), X, y, scoring='neg_mean_squared_error'\n",
        ").mean()\n",
        "baseline"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "4976.655882526393"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 124
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_LrPgQuhrvv0"
      },
      "source": [
        "def f(parameters):\n",
        "    parameters = parameters[0]\n",
        "    score = -cross_val_score(\n",
        "        SVR(gamma=parameters[2],C=parameters[0],epsilon=parameters[1]\n",
        "    ),X, y, scoring='neg_mean_squared_error').mean()\n",
        "    score = np.array(score)\n",
        "    return score"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xts6cgkSsUkF"
      },
      "source": [
        "bounds = [\n",
        "    {'name': 'C',\n",
        "     'type': 'continuous',\n",
        "     'domain': (1e-5, 1000)},\n",
        "\n",
        "    {'name': 'epsilon',\n",
        "     'type': 'continuous',\n",
        "     'domain': (1e-5, 10)},\n",
        "\n",
        "    {'name': 'gamma',\n",
        "     'type': 'continuous',\n",
        "     'domain': (1e-5, 10)},\n",
        "]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rKlWxQWctDrL"
      },
      "source": [
        "np.random.seed(777)\n",
        "optimizer = GPyOpt.methods.BayesianOptimization(\n",
        "    f=f, domain=bounds,\n",
        "    acquisition_type ='MPI',\n",
        "    acquisition_par = 0.1,\n",
        "    exact_eval=True\n",
        ")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RAlFURLYtejA"
      },
      "source": [
        "max_iter = 50\n",
        "max_time = 60\n",
        "optimizer.run_optimization(max_iter, max_time)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5MNt3sv6t6XI",
        "outputId": "92bd8d4e-6bf1-4008-9426-597b9dde0af2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 350
        }
      },
      "source": [
        "optimizer.plot_convergence()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x360 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fVpzh5yTuIFF",
        "outputId": "8b651a34-65c8-492a-e996-63083067a7ce",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "np.min(optimizer.Y)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "2911.470304239638"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 130
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ohqg4ehYuBHt"
      },
      "source": [
        "best_epsilon=optimizer.X[np.argmin(optimizer.Y)][1]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zlW_19QpLr2B",
        "outputId": "4648a65b-e124-4932-e3d4-487c5adfcf1b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "### YOUR CODE HERE\n",
        "# best_epsilon = ### YOUR CODE HERE\n",
        "grader.submit_GPyOpt_1(best_epsilon)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Current answer for task 2.1 is: 10.0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "g2N5-iAFLr2D"
      },
      "source": [
        "<b>Task 2.2:</b> For the model above submit boost in improvement that you got after tuning hyperparameters (output percents) [e.g. if baseline MSE was 40 and you got 20, output number 200]"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n09ydDjRubCF",
        "outputId": "da926221-c753-42e4-f9e5-a6e33af9f99a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "baseline = -cross_val_score(\n",
        "    SVR(gamma='auto'), X, y, scoring='neg_mean_squared_error'\n",
        ").mean()\n",
        "baseline"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "6074.97707523125"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 134
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "p-8SHaEgxx-L",
        "outputId": "3ff60da6-0983-4420-89b3-f5285d15cf39",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "baseline/np.min(optimizer.Y)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "2.0865667310379097"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 135
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "I_UYggyCLr2D",
        "outputId": "82296d61-fdc3-4fdc-b142-ef81a609045b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "performance_boost =  baseline/np.min(optimizer.Y)### YOUR CODE HERE\n",
        "grader.submit_GPyOpt_2(performance_boost*100)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Current answer for task 2.2 is: 208.65667310379098\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VSrevETiLr2G"
      },
      "source": [
        "# Authorization & Submission\n",
        "To submit assignment parts to Cousera platform, please, enter your e-mail and token into variables below. You can generate a token on this programming assignment's page. <b>Note:</b> The token expires 30 minutes after generation."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wrF8OdTPLr2G",
        "outputId": "29d8bc32-62c5-4379-ce46-15ae42c0e73a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        }
      },
      "source": [
        "STUDENT_EMAIL = \"abhishek.s.eee15@itbhu.ac.in\"# EMAIL HERE\n",
        "STUDENT_TOKEN = \"wEsHEVLhP9PgoAgJ\"# TOKEN HERE\n",
        "grader.status()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "You want to submit these numbers:\n",
            "Task 1.1: 1.0461813545396959\n",
            "Task 1.2 (mean): 0.6646774926102937\n",
            "Task 1.2 (variance): 1.1001478223790582\n",
            "Task 1.3: 1.6252681650349912\n",
            "Task 1.4 (noise): 10.143215047293623\n",
            "Task 1.4 (just signal): 3.686656201867028e-14\n",
            "Task 1.5: 4.992659262825191\n",
            "Task 2.1: 10.0\n",
            "Task 2.2: 208.65667310379098\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "86GfxlM9Lr2J"
      },
      "source": [
        "If you want to submit these answers, run cell below"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1u84ewNLLr2J",
        "outputId": "d35a3441-3c09-408b-a9a4-2e9699edaf7c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "grader.submit(STUDENT_EMAIL, STUDENT_TOKEN)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Submitted to Coursera platform. See results on assignment page!\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cjyGpKEAuy29"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}