modeling_josh.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "85d9d16b-977b-4f09-b53c-e305eba3188e",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Importing required packages and files\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from scipy import stats\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import os\n",
    "import ds_fun\n",
    "\n",
    "#These imports were constructed\n",
    "\n",
    "#Tools to build machine learning models and reports\n",
    "from sklearn.model_selection import train_test_split\n",
    "import sklearn.preprocessing\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.cluster import KMeans\n",
    "from sklearn.preprocessing import PolynomialFeatures\n",
    "from sklearn.linear_model import LinearRegression, LassoLars, TweedieRegressor\n",
    "\n",
    "#Removes warnings and imporves asthenics\n",
    "import warnings\n",
    "#warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "#Sets an option to show all columns\n",
    "pd.options.display.max_columns = None\n",
    "pd.options.display.width = 100\n",
    "pd.options.display.max_colwidth = None\n",
    "pd.options.display.max_rows = 200"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f47087e9-97d1-48c0-9e98-c4b94c714fb0",
   "metadata": {},
   "source": [
    "Acquire:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "2477c3f8-508b-46a4-b531-82b4f2449515",
   "metadata": {},
   "outputs": [],
   "source": [
    "wine_df = ds_fun.pour_wine()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ef43f523-5c22-4431-90bb-687612b34075",
   "metadata": {},
   "source": [
    "Prepare:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "f15c5a93-b51f-4790-ba11-244243db8c9a",
   "metadata": {},
   "outputs": [],
   "source": [
    "train, val, test = ds_fun.train_validate(wine_df, random_seed=1969)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "454c73b7-cdf5-4295-bb66-093506254889",
   "metadata": {},
   "outputs": [],
   "source": [
    "cont_columns = ['fixed acidity', 'volatile acidity', 'citric acid', 'residual sugar', 'chlorides',\n",
    "       'free sulfur dioxide', 'total sulfur dioxide', 'density', 'pH', 'sulphates', 'alcohol']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "e9263c30-7155-48fb-a7a9-9db6b0123576",
   "metadata": {},
   "outputs": [],
   "source": [
    "train, val, test = ds_fun.scale_cont_columns(train, val, test , cont_columns, scaler_model = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "9c0f3d55-5149-4ac6-b6b4-3148b9823051",
   "metadata": {},
   "outputs": [],
   "source": [
    "qual_class = [0,5,7,9]\n",
    "qual_bins = ['low', 'med', 'high']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "879cdf42-eb63-41d0-a64c-269ead6107e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "#train = pd.cut(qual_class, qual_bins)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bd89ca92-3375-4461-b1c6-c17e88f8fe5d",
   "metadata": {},
   "source": [
    "Engineer:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "9a38e8c6-f803-43d3-98fd-16352f1f8fe6",
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_list = [ 'density', 'volatile acidity', 'chlorides', 'alcohol']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "19e66dea-37a9-4181-92ee-d5ac5ff685d1",
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_list0 = [ 'density', 'volatile acidity']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "e2c504af-90d1-4a5e-ba40-ea6420aba85c",
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_list1 = [ 'density', 'chlorides']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "53363465-3a77-459a-a95d-728dbdeb1b87",
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_list2 = [ 'density', 'alcohol']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "6bb0f2de-283d-4902-a736-d29215eee75d",
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_list3 = [ 'volatile acidity', 'chlorides']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "3dba5e04-9435-4701-b8a6-eebcd24fd7ab",
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_list4 = ['volatile acidity', 'alcohol']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "3ff8805c-489f-4889-b5b0-e0066020c613",
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_list5 = ['chlorides', 'alcohol']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "45c1a5fc-5e80-4032-a6b4-9097de957f34",
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_list6 = ['volatile acidity', 'chlorides', 'alcohol']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "3aa46034-fbae-4312-9406-c5c0414a709f",
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_list7 = [ 'density', 'chlorides', 'alcohol']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "1cc382cc-a04a-4a47-9dc8-07a27d16f9ca",
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_list8 = [ 'density', 'volatile acidity', 'alcohol']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "81c674dd-37f7-4351-b6b4-31be552960d6",
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_list9 = [ 'density', 'volatile acidity', 'chlorides']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "682663f1-55da-40ed-8c48-41085aa44fe7",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_clusters = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "ed2c193f-4c41-4280-be51-c70894bd6bb6",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_clusters0 = 4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "d286eaaf-1595-4259-91ef-db90d1c9c8e5",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_clusters1 = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "3bf9153b-c777-42fb-aaf9-b24e5b9a25b6",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_clusters2 = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "c0554f7f-3d7d-4217-b932-37f059181f1c",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_clusters3 = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "c5decd8d-dc26-46e6-bb4c-1e552ed2a1c6",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_clusters4 = 4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "815df178-d10e-4e4d-9efb-eb67e5667942",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_clusters5 = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "74cff8fa-f010-4aa5-9859-f4da8bd3164d",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_clusters6 = 4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "6a66916b-b1ea-4f42-b079-9c3eb22eced1",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_clusters7 = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "eb6258d7-6dd8-4198-8eeb-ce3b99209006",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_clusters8 = 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "55cceee2-19e0-4635-ab7a-95cb898ea797",
   "metadata": {},
   "outputs": [],
   "source": [
    "n_clusters9 = 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "1b1b5384-a6d1-41be-a420-4ea7367a3f8b",
   "metadata": {},
   "outputs": [],
   "source": [
    "#ds_fun.elbow_method(train, cluster_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "679401aa-e441-4289-8b0e-d4f0e67d9091",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "def testing_clusters(train, val, test, cluster_list, n_clusters = 2):\n",
    "    #Builds the cluster object with the desired number of clusters\n",
    "    kmeans = KMeans(n_clusters=n_clusters)\n",
    "    kmeans.fit(train[cluster_list])\n",
    "\n",
    "    #Builds a column with the cluster numbers\n",
    "    train['cluster_preds'] = kmeans.predict(train[cluster_list])\n",
    "    val['cluster_preds'] = kmeans.predict(val[cluster_list])\n",
    "    test['cluster_preds'] = kmeans.predict(test[cluster_list])\n",
    "\n",
    "\n",
    "    train.drop(columns=cluster_list, inplace = True)\n",
    "    val.drop(columns=cluster_list, inplace = True) \n",
    "    test.drop(columns=cluster_list, inplace = True) \n",
    "          \n",
    "    dumb_list = ['type', 'cluster_preds']\n",
    "\n",
    "    \n",
    "    train = pd.get_dummies(train, columns = dumb_list)\n",
    "\n",
    "    val = pd.get_dummies(val, columns = dumb_list)\n",
    "\n",
    "    test = pd.get_dummies(test, columns = dumb_list)\n",
    "\n",
    "    return train, val, test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "25e05acc-296a-41bc-ad4e-34a925be26e3",
   "metadata": {},
   "outputs": [],
   "source": [
    "#train, val, test = testing_clusters(train, val, test, cluster_list, n_clusters = n_clusters)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "51934610-03a6-4841-a08c-ee2f0595aa3b",
   "metadata": {},
   "source": [
    "kmeans = KMeans(n_clusters=n_clusters)\n",
    "kmeans.fit(train[cluster_list])\n",
    "\n",
    "#Builds a column with the cluster numbers\n",
    "train['cluster_preds'] = kmeans.predict(train[cluster_list])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "26b9ea97-9c00-4318-90e3-e5dcf7525a6b",
   "metadata": {},
   "source": [
    "pd.get_dummies(train,columns=['type', 'cluster_preds'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6fbb6e6f-28db-4ebe-88fb-5738e96da3af",
   "metadata": {},
   "source": [
    "dummy_df = pd.get_dummies(train[['cluster_preds', 'type']])\n",
    "dummy_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "16b9900c-264f-4f0e-93ba-0dbd77bdbb6b",
   "metadata": {},
   "outputs": [],
   "source": [
    "train = pd.get_dummies(train, columns=['type'])\n",
    "val = pd.get_dummies(val, columns=['type'])\n",
    "test = pd.get_dummies(test, columns=['type'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "b9d8352d-c383-4b76-be7e-07e2f6623c51",
   "metadata": {},
   "outputs": [],
   "source": [
    "target_col = 'quality'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "3b224801-4843-4059-92ae-ff73f4f46948",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, y_train, X_val, y_val, X_test, y_test = ds_fun.train_val_test(train, val, test, target_col)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "a5414f44-6a8b-4249-b92f-1fa8b26e411e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Using RMSE Mean training baseline: 0.8640\n"
     ]
    }
   ],
   "source": [
    "baseline = ds_fun.find_regression_baseline(y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "90079568-9c38-4cee-84a0-6dcc0423df8d",
   "metadata": {},
   "outputs": [],
   "source": [
    "#List for gathering metrics\n",
    "rmse_scores = []\n",
    "\n",
    "\n",
    "\"\"\" *** Builds and fits Linear Regression Model (OLS) *** \"\"\"\n",
    "\n",
    "\n",
    "lm = LinearRegression(normalize=True, positive=True)\n",
    "lm.fit(X_train, y_train)\n",
    "\n",
    "#Train data\n",
    "lm_preds = pd.DataFrame({'actual':y_train})\n",
    "lm_preds['pred_lm'] = lm.predict(X_train)\n",
    "\n",
    "#Validate data\n",
    "lm_val_preds = pd.DataFrame({'actual':y_val})\n",
    "lm_val_preds['lm_val_preds'] = lm.predict(X_val)\n",
    "\n",
    "#Finds score on Train and Validate data\n",
    "rmse_train = mean_squared_error(lm_preds['actual'],\n",
    "                                lm_preds['pred_lm'],\n",
    "                                squared=False) \n",
    "rmse_val = mean_squared_error(lm_val_preds['actual'],\n",
    "                              lm_val_preds['lm_val_preds'],\n",
    "                              squared=False) \n",
    "\n",
    "#Adds score to metrics list for later comparison\n",
    "rmse_scores.append({'Model':'OLS Linear',\n",
    "                'RMSE on Train': round(rmse_train,4),\n",
    "                'RMSE on Validate': round(rmse_val,4)})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "894cb797-dd17-49e8-98dd-89da63500fe9",
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" *** Builds and fits Lasso Lars Model *** \"\"\"\n",
    "\n",
    "\n",
    "lars = LassoLars(alpha=.25, max_iter=1000) #\n",
    "lars.fit(X_train, y_train)\n",
    "\n",
    "#Train data\n",
    "ll_preds = pd.DataFrame({'actual':y_train})\n",
    "ll_preds['pred_ll'] = lars.predict(X_train)\n",
    "\n",
    "#Validate data\n",
    "ll_val_preds = pd.DataFrame({'actual':y_val})\n",
    "ll_val_preds['ll_val_preds'] = lars.predict(X_val)\n",
    "\n",
    "#Finds score on Train and Validate data\n",
    "rmse_train = mean_squared_error(ll_preds['actual'],\n",
    "                                ll_preds['pred_ll'],\n",
    "                                squared=False)\n",
    "rmse_val = mean_squared_error(ll_val_preds['actual'],\n",
    "                              ll_val_preds['ll_val_preds'],\n",
    "                              squared=False)\n",
    "\n",
    "#Adds score to metrics list for later comparison\n",
    "rmse_scores.append({'Model':'Lasso Lars',\n",
    "                'RMSE on Train': round(rmse_train,4),\n",
    "                'RMSE on Validate': round(rmse_val,4)})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "b19f5041-01d8-4c0e-90aa-8cb72decf769",
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" *** Builds and fits Tweedie Regressor (GLM) Model *** \"\"\"\n",
    "\n",
    "glm = TweedieRegressor(power=1, alpha=1)    \n",
    "glm.fit(X_train, y_train)\n",
    "\n",
    "#Train data\n",
    "glm_preds = pd.DataFrame({'actual':y_train})\n",
    "glm_preds['pred_glm'] = glm.predict(X_train)\n",
    "\n",
    "#Validate data\n",
    "glm_val_preds = pd.DataFrame({'actual':y_val})\n",
    "glm_val_preds['glm_val_preds'] = glm.predict(X_val)\n",
    "\n",
    "#Finds score on Train and Validate data\n",
    "rmse_train = mean_squared_error(glm_preds['actual'],\n",
    "                                glm_preds['pred_glm'],\n",
    "                                squared=False) \n",
    "rmse_val = mean_squared_error(glm_val_preds['actual'],\n",
    "                              glm_val_preds['glm_val_preds'],\n",
    "                              squared=False)\n",
    "\n",
    "#Adds score to metrics list for later comparison\n",
    "rmse_scores.append({'Model':'Tweedie',\n",
    "                    'RMSE on Train': round(rmse_train,4),\n",
    "                    'RMSE on Validate': round(rmse_val,4)})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "6a158a02-5eb3-47eb-869b-79acf292fc2b",
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" *** Builds and fits Polynomial regression Model *** \"\"\"\n",
    "\n",
    "\n",
    "#Polynomial Regression part:\n",
    "# make the polynomial features to get a new set of features\n",
    "pf = PolynomialFeatures(degree=1)\n",
    "\n",
    "# fit and transform X_train_scaled\n",
    "X_train_degree2 = pf.fit_transform(X_train)\n",
    "\n",
    "# transform X_validate_scaled & X_test_scaled\n",
    "X_validate_degree2 = pf.transform(X_val)\n",
    "#X_test_degree2 = pf.transform(X_test)\n",
    "\n",
    "#Polynomial Regression being fed into Linear Regression:\n",
    "lm2 = LinearRegression(normalize=True)\n",
    "lm2.fit(X_train_degree2, y_train)\n",
    "\n",
    "#Train data\n",
    "lm2_preds = pd.DataFrame({'actual':y_train})\n",
    "lm2_preds['pred_lm2'] = lm2.predict(X_train_degree2)\n",
    "\n",
    "#Validate data\n",
    "lm2_val_preds = pd.DataFrame({'actual':y_val})\n",
    "lm2_val_preds['lm2_val_preds'] = lm2.predict(X_validate_degree2)\n",
    "\n",
    "#Finds score on Train and Validate data\n",
    "rmse_train = mean_squared_error(lm2_preds['actual'],\n",
    "                                lm2_preds['pred_lm2'],\n",
    "                                squared=False) \n",
    "rmse_val = mean_squared_error(lm2_val_preds['actual'],\n",
    "                              lm2_val_preds['lm2_val_preds'],\n",
    "                              squared=False)\n",
    "\n",
    "#Adds score to metrics list for later comparison\n",
    "rmse_scores.append({'Model':'Polynomial',\n",
    "                    'RMSE on Train': round(rmse_train,4),\n",
    "                    'RMSE on Validate': round(rmse_val,4)})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "c013e09b-4a7c-4098-b854-764ebb139f37",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "        Model  RMSE on Train  RMSE on Validate  Difference\n",
      "0  OLS Linear         0.7130            0.7511       -0.04\n",
      "1  Lasso Lars         0.8640            0.8876       -0.02\n",
      "2     Tweedie         0.8182            0.8406       -0.02\n",
      "3  Polynomial         0.7130            0.7511       -0.04\n"
     ]
    }
   ],
   "source": [
    "\"\"\" *** Later comparison section to display results *** \"\"\"\n",
    "\n",
    "#Builds and displays results DataFrame\n",
    "rmse_scores = pd.DataFrame(rmse_scores)\n",
    "rmse_scores['Difference'] = round(rmse_scores['RMSE on Train'] - rmse_scores['RMSE on Validate'],2)    \n",
    "\n",
    "#Results were too close so had to look at the numbers\n",
    "print(rmse_scores)\n",
    "\n",
    "#Building variables for plotting\n",
    "rmse_min = min([rmse_scores['RMSE on Train'].min(),\n",
    "                rmse_scores['RMSE on Validate'].min(), baseline])\n",
    "rmse_max = max([rmse_scores['RMSE on Train'].max(),\n",
    "                rmse_scores['RMSE on Validate'].max(), baseline])\n",
    "\n",
    "lower_limit = rmse_min * 0.8\n",
    "upper_limit = rmse_max * 1.05"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "a85dba30-7b06-4083-8206-28902c67b2a0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "x = np.arange(len(rmse_scores))  # the label locations\n",
    "width = 0.25  # the width of the bars\n",
    "\n",
    "fig, ax = plt.subplots(facecolor=\"gainsboro\")\n",
    "rects1 = ax.bar(x - width/2, rmse_scores['RMSE on Train'],\n",
    "                width, label='Training data', color='#4e5e33',\n",
    "                edgecolor='dimgray') #Codeup dark green\n",
    "rects2 = ax.bar(x + width/2, rmse_scores['RMSE on Validate'],\n",
    "                width, label='Validation data', color='#8bc34b',\n",
    "                edgecolor='dimgray') #Codeup light green\n",
    "\n",
    "# Need to have baseline input:\n",
    "plt.axhline(baseline, label=\"Baseline Error\", c='tomato', linestyle=':')\n",
    "\n",
    "# Add some text for labels, title and custom x-axis tick labels, etc.\n",
    "ax.axhspan(0, baseline, facecolor='palegreen', alpha=0.2)\n",
    "ax.axhspan(baseline, upper_limit, facecolor='red', alpha=0.3)\n",
    "ax.set_ylabel('RMS Error')\n",
    "ax.set_xlabel('Machine Learning Models')\n",
    "ax.set_title('Model Error Scores')\n",
    "ax.set_xticks(x, rmse_scores['Model'])\n",
    "\n",
    "plt.ylim(bottom=lower_limit, top = upper_limit)\n",
    "\n",
    "ax.legend(loc='upper right', framealpha=.9, facecolor=\"whitesmoke\",\n",
    "          edgecolor='darkolivegreen')\n",
    "\n",
    "fig.tight_layout()\n",
    "#plt.savefig('best_model_all_features.png')\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}