merge_performance.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Intro\n",
    "Benchmarking the performance of merges. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas\n",
    "import numpy\n",
    "import matplotlib.pyplot as plt\n",
    "import timeit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('0.25.3', '1.17.3')"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pandas.__version__, numpy.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def generate_dataframes(N:int, set_index: bool, type_def: str, duplicates:float):\n",
    "    \n",
    "    numpy.random.seed(11)\n",
    "    if duplicates >0:\n",
    "        sample = int((1-duplicates)*N)+1\n",
    "        array_1 = numpy.random.choice(sample, N, replace=True)\n",
    "    else:\n",
    "        array_1 = numpy.arange(N)\n",
    "        numpy.random.shuffle(array_1) # shuffle works in-place\n",
    "        \n",
    "    df_1 = pandas.DataFrame(array_1, columns=[\"A\"])\n",
    "    df_1[\"B\"] = df_1[\"A\"]*2\n",
    "\n",
    "    array_2 = array_1.copy()\n",
    "    numpy.random.shuffle(array_2)\n",
    "    df_2 = pandas.DataFrame(array_2, columns=[\"A\"])\n",
    "    df_2[\"B\"] = df_2[\"A\"]*2\n",
    "    \n",
    "    if type_def == \"datetime\":\n",
    "        df_1[\"A\"] = pandas.to_datetime(df_1[\"A\"])\n",
    "        df_2[\"A\"] = pandas.to_datetime(df_2[\"A\"])\n",
    "    else:\n",
    "        df_1[\"A\"] =df_1[\"A\"].astype(type_def)\n",
    "        df_2[\"A\"] =df_2[\"A\"].astype(type_def)\n",
    "    \n",
    "    if set_index:\n",
    "        df_1 = df_1.set_index(\"A\")\n",
    "        df_2 = df_2.set_index(\"A\")\n",
    "    \n",
    "    return df_1, df_2\n",
    "\n",
    "def get_time(df_1: pandas.DataFrame, df_2: pandas.DataFrame, indexed: bool):\n",
    "    if indexed:\n",
    "        time_result = %timeit -o df_1.merge(df_2, how=\"left\", left_index=True, right_index=True)\n",
    "    else:\n",
    "        time_result = %timeit -o df_1.merge(df_2, how=\"left\", left_on=\"A\", right_on=\"A\")\n",
    "    return time_result.average\n",
    "\n",
    "def profile(df_1: pandas.DataFrame, df_2: pandas.DataFrame, indexed: bool):\n",
    "    if indexed:\n",
    "        df_1.merge(df_2, how=\"left\", left_index=True, right_index=True)\n",
    "    else:\n",
    "        df_1.merge(df_2, how=\"left\", left_on=\"A\", right_on=\"A\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.78 ms ± 105 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "1.85 ms ± 139 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n",
      "1.77 ms ± 38.3 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "1.92 ms ± 37.4 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "3.07 ms ± 23.3 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n",
      "28.6 ms ± 420 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "504 ms ± 5.36 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "6.98 s ± 113 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "1min 28s ± 2.99 s per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "430 µs ± 3.94 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "534 µs ± 8.34 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "532 µs ± 3.14 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "564 µs ± 21.1 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "708 µs ± 35.8 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "2.46 ms ± 96.5 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n",
      "67.7 ms ± 805 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "1.01 s ± 9.88 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "16.6 s ± 150 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "#1. Compare performance of merge for various sizes when joining on index vs not on index\n",
    "sizes = [1,10,100,1_000, 10_000, 100_000, 1_000_000, 10_000_000, 100_000_000] \n",
    "\n",
    "dfs = [generate_dataframes(N, False, \"int\", duplicates=0) for N in sizes]\n",
    "time_by_size = [get_time(df_1, df_2, False) for (df_1, df_2) in dfs]\n",
    "\n",
    "dfs_indexed = [generate_dataframes(N, True, \"int\", duplicates=0) for N in sizes]\n",
    "time_by_size_indexed = [get_time(df_1, df_2, True) for (df_1, df_2) in dfs_indexed]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot results\n",
    "fig, ax = plt.subplots()\n",
    "ax.plot(sizes, time_by_size, 'o', label=\"not on index\")\n",
    "ax.plot(sizes, time_by_size_indexed, 'o', label=\"on index\")\n",
    "ax.set_xlabel(\"number of rows\")\n",
    "ax.set_ylabel(\"time (s)\")\n",
    "ax.set_xscale(\"log\")\n",
    "ax.set_yscale(\"log\")\n",
    "ax.legend()\n",
    "fig.savefig(\"merge-performance-by-size.png\")\n",
    "fig.savefig(\"merge-performance-by-size.svg\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "508 ms ± 4.64 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "850 ms ± 20 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "499 ms ± 3.52 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "1.2 s ± 24.9 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "652 ms ± 32 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "998 ms ± 11.6 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "58.2 ms ± 556 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "223 ms ± 4.34 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "70.5 ms ± 644 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "336 ms ± 2.37 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "2.28 s ± 13.3 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "82.8 ms ± 2.22 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "#2. Compare performance of merging on different dtypes (on index vs not on index)\n",
    "type_defs = [\"datetime\", pandas.Int64Dtype(), \"int\", \"str\", \"category\", \"float\"]\n",
    "dfs = [generate_dataframes(1_000_000, False, type_def, duplicates=0) for type_def in type_defs]\n",
    "time_by_type = [get_time(df_1, df_2, False) for (df_1, df_2) in dfs]\n",
    "\n",
    "dfs = [generate_dataframes(1_000_000, True, type_def, duplicates=0) for type_def in type_defs]\n",
    "time_by_type_indexed = [get_time(df_1, df_2, True) for (df_1, df_2) in dfs]\n",
    "type_defs = [\"datetime\", \"nullable-int\", \"int\", \"str\", \"category\", \"float\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "ax.plot(type_defs, time_by_type, 'o', label=\"not on index\")\n",
    "ax.plot(type_defs, time_by_type_indexed, 'o', label=\"on index\")\n",
    "ax.set_xlabel(\"dtype\")\n",
    "ax.set_ylabel(\"time (s)\")\n",
    "ax.set_yscale(\"log\")\n",
    "ax.legend()\n",
    "fig.savefig(\"merge-performance-by-type.png\")\n",
    "fig.savefig(\"merge-performance-by-type.svg\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "514 ms ± 14.9 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "583 ms ± 17.5 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "600 ms ± 22.3 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "611 ms ± 15 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "583 ms ± 8.42 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "663 ms ± 15.7 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "940 ms ± 99.8 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "1.39 s ± 105 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "2.28 s ± 164 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "70.8 ms ± 3.63 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "965 ms ± 103 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "999 ms ± 95.3 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "976 ms ± 76.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "953 ms ± 43.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "789 ms ± 83.1 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "812 ms ± 41.5 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "1.01 s ± 27.6 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "1.47 s ± 75.1 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "#3. performance depending on duplicate amount (proportion)\n",
    "duplicates = [0,0.05, 0.1, 0.2,0.4,0.6, 0.8, 0.9, 0.95]\n",
    "\n",
    "dfs = [generate_dataframes(1_000_000, False, \"int\", duplicates=d) for d in duplicates]\n",
    "time_by_duplicates = [get_time(df_1, df_2, False) for (df_1, df_2) in dfs]\n",
    "\n",
    "dfs_indexed = [generate_dataframes(1_000_000, True, \"int\", duplicates=d) for d in duplicates]\n",
    "time_by_duplicates_indexed = [get_time(df_1, df_2, True) for (df_1, df_2) in dfs_indexed]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "ax.plot(duplicates, time_by_duplicates, 'o', label=\"not on index\")\n",
    "ax.plot(duplicates, time_by_duplicates_indexed, 'o', label=\"on index\")\n",
    "ax.set_xlabel(\"duplicate density\")\n",
    "ax.set_ylabel(\"time (s)\")\n",
    "ax.set_yscale(\"log\")\n",
    "ax.legend()\n",
    "fig.savefig(\"merge-performance-by-duplicate.png\")\n",
    "fig.savefig(\"merge-performance-by-duplicate.svg\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "66.8 ms ± 2.06 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.06684142000001105"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#4. Sorted vs unsorted\n",
    "# (unsorted)\n",
    "df_1, df_2 = generate_dataframes(1_000_000, True, \"int\", duplicates=0)\n",
    "get_time(df_1, df_2, True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "7.02 ms ± 343 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.007017911428570838"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# sorted dfs merge much faster...\n",
    "df_1, df_2 = generate_dataframes(1_000_000, True, \"int\", duplicates=0)\n",
    "df_1 = df_1.sort_values(\"A\", ascending=True)\n",
    "df_2 = df_2.sort_values(\"A\", ascending=True)\n",
    "get_time(df_1, df_2, True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "44.9 ms ± 1.73 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.04490464285714292"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_1, df_2 = generate_dataframes(1_000_000, True, \"int\", duplicates=0)\n",
    "df_1 = df_1.sort_values(\"A\", ascending=True)\n",
    "get_time(df_1, df_2, True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.83 ms ± 100 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "1.97 ms ± 141 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n",
      "1.94 ms ± 35.2 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "1.99 ms ± 124 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n",
      "3.22 ms ± 84.2 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n",
      "35.4 ms ± 8.64 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "545 ms ± 29.3 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "7.71 s ± 552 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "1.76 ms ± 84.4 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "1.78 ms ± 72.2 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "1.95 ms ± 197 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "2.03 ms ± 294 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n",
      "2.88 ms ± 54.8 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n",
      "27.8 ms ± 1.34 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n",
      "414 ms ± 22.6 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "4.65 s ± 56.9 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "434 µs ± 951 ns per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "542 µs ± 6.17 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "556 µs ± 6.6 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "571 µs ± 11.1 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "747 µs ± 66.2 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "2.77 ms ± 581 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n",
      "108 ms ± 45.8 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "858 ms ± 10.5 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n",
      "436 µs ± 1.91 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "444 µs ± 6.64 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "439 µs ± 5.68 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "450 µs ± 16.4 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "467 µs ± 5.79 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "654 µs ± 54.3 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n",
      "6.84 ms ± 155 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n",
      "80.6 ms ± 14.2 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "sizes = [1,10,100,1_000, 10_000, 100_000, 1_000_000, 10_000_000] \n",
    "\n",
    "dfs = [generate_dataframes(N, False, \"int\", duplicates=0) for N in sizes]\n",
    "time_by_size = [get_time(df_1, df_2, False) for (df_1, df_2) in dfs]\n",
    "\n",
    "dfs_sorted = [(df_1.copy().sort_values(\"A\", ascending=True),\n",
    "               df_2.copy().sort_values(\"A\", ascending=True)) for (df_1, df_2) in dfs]\n",
    "time_by_size_sorted = [get_time(df_1, df_2, False) for (df_1, df_2) in dfs_sorted]\n",
    "\n",
    "dfs_indexed = [generate_dataframes(N, True, \"int\", duplicates=0) for N in sizes]\n",
    "time_by_size_indexed = [get_time(df_1, df_2, True) for (df_1, df_2) in dfs_indexed]\n",
    "\n",
    "dfs_indexed_sorted = [(df_1.copy().sort_values(\"A\", ascending=True),\n",
    "                       df_2.copy().sort_values(\"A\", ascending=True)) for (df_1, df_2) in dfs_indexed]\n",
    "time_by_size_indexed_sorted = [get_time(df_1, df_2, True) for (df_1, df_2) in dfs_indexed_sorted]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots()\n",
    "ax.plot(sizes, time_by_size, 'o', label=\"not on index (not sorted)\")\n",
    "ax.plot(sizes, time_by_size_sorted, 'o', label=\"not on index (sorted)\")\n",
    "ax.plot(sizes, time_by_size_indexed, 'o', label=\"on index (not sorted)\")\n",
    "ax.plot(sizes, time_by_size_indexed_sorted, 'o', label=\"on index (sorted)\")\n",
    "ax.set_xlabel(\"number of rows\")\n",
    "ax.set_ylabel(\"time (s)\")\n",
    "ax.set_xscale(\"log\")\n",
    "ax.set_yscale(\"log\")\n",
    "ax.legend()\n",
    "fig.savefig(\"merge-performance-by-size-sorting.png\")\n",
    "fig.savefig(\"merge-performance-by-size-sorting.svg\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# map vs merge - for small mapping tables using map is faster than join\n",
    "df_1, df_2 = generate_dataframes(1_000_000, False, \"int\", duplicates=0.999)\n",
    "df_3 = pandas.DataFrame([x for x in range(df_1[\"A\"].max()+1)], columns=[\"A\"])\n",
    "df_3[\"B\"]=[x*2 for x in range(df_1[\"A\"].max()+1)]\n",
    "mapping ={x:x*2 for x in range(df_1[\"A\"].max()+1)}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "get_time(df_1, df_3, False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%timeit df_1[\"A\"].map(mapping)"
   ]
  }
 ],
 "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": "3.7.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}