XGboost classifier with hyperparameter tuning.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "20140930",
   "metadata": {},
   "source": [
    "# Import the libraries needed\n",
    "These should be specific to the task at hand which in this case is hyperparameter tuning and classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "285dae48",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Success - the MySageMakerInstance is in the us-west-1 region. You will use the 632365934929.dkr.ecr.us-west-1.amazonaws.com/xgboost:latest container for your SageMaker endpoint.\n"
     ]
    }
   ],
   "source": [
    "#importing necessary libraries\n",
    "import boto3, re, sys, math, json, os, sagemaker, urllib.request\n",
    "from sagemaker import get_execution_role\n",
    "\n",
    "from IPython.display import Image\n",
    "from IPython.display import display\n",
    "from time import gmtime, strftime\n",
    "from sagemaker.predictor import csv_serializer\n",
    "from sagemaker.tuner import (\n",
    "    IntegerParameter,\n",
    "    CategoricalParameter,\n",
    "    ContinuousParameter,\n",
    "    HyperparameterTuner,\n",
    ")\n",
    "from sagemaker.serializers import CSVSerializer\n",
    "from sagemaker.amazon.amazon_estimator import get_image_uri\n",
    "from sagemaker.image_uris import retrieve\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import auc, accuracy_score, confusion_matrix, mean_squared_error\n",
    "from sklearn.metrics import accuracy_score,classification_report\n",
    "import seaborn as sns\n",
    "\n",
    "smclient = boto3.Session().client('sagemaker')\n",
    "# Define IAM role\n",
    "role = get_execution_role()\n",
    "prefix = '###################'\n",
    "my_region = boto3.session.Session().region_name \n",
    "\n",
    "xgboost_container = sagemaker.image_uris.retrieve(\"xgboost\", my_region, \"latest\")\n",
    "\n",
    "print(\"Success - the MySageMakerInstance is in the \" + my_region + \n",
    "      \" region. You will use the \" + xgboost_container + \n",
    "      \" container for your SageMaker endpoint.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5fddd874",
   "metadata": {},
   "source": [
    "# Create bucket if not pre-existing\n",
    "In this case it was created already"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "5d87da57",
   "metadata": {},
   "outputs": [],
   "source": [
    "##name of pre-existing bucket\n",
    "bucket_name = '#################' "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a73dd4f9",
   "metadata": {},
   "source": [
    "# Upload the data to the bucket\n",
    "Data was already explicitly uploaded to our pre-existing bucket"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "5089b5e1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Success: Data loaded into dataframe.\n"
     ]
    }
   ],
   "source": [
    "try:\n",
    "    model_data = pd.read_csv('###########.csv',index_col=0)\n",
    "    print('Success: Data loaded into dataframe.')\n",
    "except Exception as e:\n",
    "    print('Data load error: ',e)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "378485f3",
   "metadata": {},
   "source": [
    "# Standard train/test split of 70/30"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "da63a164",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(3476, 7) (1490, 7)\n"
     ]
    }
   ],
   "source": [
    "train_data, test_data = np.split(model_data.sample(frac=1, random_state=1729), [int(0.7 * len(model_data))])\n",
    "print(train_data.shape, test_data.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "94d51e45",
   "metadata": {},
   "source": [
    "# Visualize possible correlations\n",
    "Possible feature engineering from data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "74d9e726",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "##heatmap for correlation\n",
    "corrMatrix = model_data.corr()\n",
    "sns.heatmap(corrMatrix, annot=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a640c262",
   "metadata": {},
   "source": [
    "# Data in proper formats for tuning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "c86f13c7",
   "metadata": {},
   "outputs": [],
   "source": [
    "##train set for hyperparameter tuning\n",
    "pd.concat([train_data['fund_stat'], train_data.drop(['fund_stat'], axis=1)], axis=1).to_csv('train.csv', index=False, header=False)\n",
    "boto3.Session().resource('s3').Bucket(bucket_name).Object(os.path.join(prefix, 'train/train.csv')).upload_file('train.csv')\n",
    "s3_input_train = sagemaker.inputs.TrainingInput(s3_data='s3://{}/{}/train'.format(bucket_name, prefix), content_type='csv')\n",
    "##validation test set for hyperparameter tuning\n",
    "pd.concat([test_data['fund_stat'], test_data.drop(['fund_stat'], axis=1)], axis=1).to_csv('test.csv', index=False, header=False)\n",
    "boto3.Session().resource('s3').Bucket(bucket_name).Object(os.path.join(prefix, 'test/test.csv')).upload_file('test.csv')\n",
    "s3_input_validation=sagemaker.inputs.TrainingInput(s3_data='s3://{}/{}/test'.format(bucket_name, prefix), content_type='csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f7c45d36",
   "metadata": {},
   "source": [
    "# Create session"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "35e2e76b",
   "metadata": {},
   "outputs": [],
   "source": [
    "sess = sagemaker.Session()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "27e39127",
   "metadata": {},
   "source": [
    "# Set your estimator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "d3b234f5",
   "metadata": {},
   "outputs": [],
   "source": [
    "xgb = sagemaker.estimator.Estimator(\n",
    "    xgboost_container,\n",
    "    role,\n",
    "    base_job_name=\"xgboost-random-search\",\n",
    "    instance_count=1,\n",
    "    instance_type=\"ml.m4.xlarge\",\n",
    "    output_path=\"s3://{}/{}/output\".format(bucket_name, prefix),\n",
    "    sagemaker_session=sess,\n",
    ")\n",
    "\n",
    "xgb.set_hyperparameters(\n",
    "    eval_metric=\"auc\",\n",
    "    objective=\"binary:logistic\",\n",
    "    num_round=10,\n",
    "    rate_drop=0.3,\n",
    "    tweedie_variance_power=1.4,\n",
    ")\n",
    "objective_metric_name = \"validation:auc\""
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bec482e0",
   "metadata": {},
   "source": [
    "# Logarithmic tuner\n",
    "You are able to set the parameters to be tuned"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "85855bb5",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "No finished training job found associated with this estimator. Please make sure this estimator is only used for building workflow config\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "....................................................................!\n"
     ]
    }
   ],
   "source": [
    "##log tuner\n",
    "hyperparameter_ranges = {\n",
    "    \"alpha\": ContinuousParameter(0.01, 10, scaling_type=\"Logarithmic\"),\n",
    "    \"lambda\": ContinuousParameter(0.01, 10, scaling_type=\"Logarithmic\"),\n",
    "}\n",
    "\n",
    "tuner_log = HyperparameterTuner(\n",
    "    xgb,\n",
    "    objective_metric_name,\n",
    "    hyperparameter_ranges,\n",
    "    max_jobs=5,\n",
    "    max_parallel_jobs=5,\n",
    "    strategy=\"Random\",\n",
    ")\n",
    "\n",
    "tuner_log.fit(\n",
    "    {\"train\": s3_input_train, \"validation\": s3_input_validation},\n",
    "    include_cls_metadata=False,\n",
    "    job_name=\"xgb-randsearch-\" + strftime(\"%Y%m%d-%H-%M-%S\", gmtime()),\n",
    ")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "13cd6c52",
   "metadata": {},
   "source": [
    "# Check the progress"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "1964035f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Completed'"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "boto3.client(\"sagemaker\").describe_hyper_parameter_tuning_job(\n",
    "    HyperParameterTuningJobName=tuner_log.latest_tuning_job.job_name\n",
    ")[\"HyperParameterTuningJobStatus\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1d83a42d",
   "metadata": {},
   "source": [
    "# Linear tuner\n",
    "It's good to check both to find the best possible model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "a3703180",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "No finished training job found associated with this estimator. Please make sure this estimator is only used for building workflow config\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "..............................................................!\n"
     ]
    }
   ],
   "source": [
    "##linear tuner\n",
    "hyperparameter_ranges_linear = {\n",
    "    \"alpha\": ContinuousParameter(0.01, 10, scaling_type=\"Linear\"),\n",
    "    \"lambda\": ContinuousParameter(0.01, 10, scaling_type=\"Linear\"),\n",
    "}\n",
    "tuner_linear = HyperparameterTuner(\n",
    "    xgb,\n",
    "    objective_metric_name,\n",
    "    hyperparameter_ranges_linear,\n",
    "    max_jobs=5,\n",
    "    max_parallel_jobs=5,\n",
    "    strategy=\"Random\",\n",
    ")\n",
    "\n",
    "tuner_linear.fit(\n",
    "    {\"train\": s3_input_train, \"validation\": s3_input_validation},\n",
    "    include_cls_metadata=False,\n",
    "    job_name=\"xgb-linsearch-\" + strftime(\"%Y%m%d-%H-%M-%S\", gmtime()),\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d40ef754",
   "metadata": {},
   "source": [
    "# Check progress"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "1d8eb834",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Completed'"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "boto3.client(\"sagemaker\").describe_hyper_parameter_tuning_job(\n",
    "    HyperParameterTuningJobName=tuner_linear.latest_tuning_job.job_name\n",
    ")[\"HyperParameterTuningJobStatus\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d3dc3762",
   "metadata": {},
   "source": [
    "# Completing status of tunings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "92ea740e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# check jobs have finished\n",
    "status_log = boto3.client(\"sagemaker\").describe_hyper_parameter_tuning_job(\n",
    "    HyperParameterTuningJobName=tuner_log.latest_tuning_job.job_name\n",
    ")[\"HyperParameterTuningJobStatus\"]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "33fba73e",
   "metadata": {},
   "outputs": [],
   "source": [
    "status_linear = boto3.client(\"sagemaker\").describe_hyper_parameter_tuning_job(\n",
    "    HyperParameterTuningJobName=tuner_linear.latest_tuning_job.job_name\n",
    ")[\"HyperParameterTuningJobStatus\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "7f72b8e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "assert status_log == \"Completed\", \"First must be completed, was {}\".format(status_log)\n",
    "assert status_linear == \"Completed\", \"Second must be completed, was {}\".format(status_linear)\n",
    "\n",
    "df_log = sagemaker.HyperparameterTuningJobAnalytics(\n",
    "    tuner_log.latest_tuning_job.job_name\n",
    ").dataframe()\n",
    "df_linear = sagemaker.HyperparameterTuningJobAnalytics(\n",
    "    tuner_linear.latest_tuning_job.job_name\n",
    ").dataframe()\n",
    "df_log[\"scaling\"] = \"log\"\n",
    "df_linear[\"scaling\"] = \"linear\"\n",
    "df = pd.concat([df_log, df_linear], ignore_index=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "221505c2",
   "metadata": {},
   "source": [
    "# Visualization of the scaling distribution of both models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "92086095",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x216 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "g = sns.FacetGrid(df, col=\"scaling\", palette=\"viridis\")\n",
    "g = g.map(plt.scatter, \"alpha\", \"lambda\", alpha=0.6)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "461a53f3",
   "metadata": {},
   "source": [
    "# Deploy the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "b8e704fe",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "2022-10-19 15:14:54 Starting - Preparing the instances for training\n",
      "2022-10-19 15:14:54 Downloading - Downloading input data\n",
      "2022-10-19 15:14:54 Training - Training image download completed. Training in progress.\n",
      "2022-10-19 15:14:54 Uploading - Uploading generated training model\n",
      "2022-10-19 15:14:54 Completed - Resource retained for reuse\n",
      "------!"
     ]
    }
   ],
   "source": [
    "predictor = tuner_linear.deploy(initial_instance_count=1, instance_type=\"ml.m4.xlarge\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d58be662",
   "metadata": {},
   "source": [
    "# Endpoint name\n",
    "Look for this in the bucket"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "a8dc5eb9",
   "metadata": {},
   "outputs": [],
   "source": [
    "##predictor.endpoint_name"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "36a7f6c7",
   "metadata": {},
   "source": [
    "# Get the results in arrays for visualization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "f1bbcca1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1490,)\n"
     ]
    }
   ],
   "source": [
    "test_target=test_data['fund_stat']\n",
    "test_data_array = test_data.drop(['fund_stat'], axis=1).values #load the data into an array\n",
    "predictor.serializer = CSVSerializer() # set the serializer type\n",
    "predictions = predictor.predict(test_data_array).decode('utf-8') # predict!\n",
    "predictions_array = np.fromstring(predictions[1:], sep=',') # and turn the prediction into an array\n",
    "print(predictions_array.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "481df37e",
   "metadata": {},
   "source": [
    "# Confusion matrix with accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "d88ff408",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 257.44, 'Predicted label')"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAagAAAFBCAYAAAAi+TuKAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8/fFQqAAAACXBIWXMAAAsTAAALEwEAmpwYAAAiGklEQVR4nO3de5xVVf3/8dd7BkW8YFwEkYuigYSUeYm8pPnVTEoNrSzUkgwjzaQ0Uyh/mvalLPtWat7wSqkIZiZpokYi3lFJRUSFRGEEQQEVjRDw8/tj78HjOMzMOTNnzmbP++ljP9hn7bX3WnvgMR/X5ayliMDMzCxrqipdATMzs/o4QJmZWSY5QJmZWSY5QJmZWSY5QJmZWSY5QJmZWSY5QFmmSeog6W+S3pR0czOec6yku1uybpUiaT9Jz1e6HmblJn8PylqCpGOA04ABwErgSWBsRDzQzOd+EzgF2Cci1ja3nlknKYB+ETGv0nUxqzS3oKzZJJ0G/B74BdAd6ANcCgxtgcdvD7zQFoJTU0hqV+k6mLUWByhrFklbA+cBJ0fEXyLinYhYExF/i4gfp3naS/q9pEXp8XtJ7dNrB0iqkfQjSUslLZZ0fHrtXOBs4OuS3pY0QtLPJF1fUP4OkqL2F7ekb0l6UdJKSfMlHVuQ/kDBfftIeiztOnxM0j4F16ZJ+rmkB9Pn3C2p6wbev7b+ZxTU/whJX5T0gqTlkn5SkH+wpIclvZHm/YOkTdNr09NsT6Xv+/WC558p6VXg2tq09J6d0jJ2Tz9vJ+l1SQc05+/VLAscoKy59gY2A25tIM9Pgb2ATwK7AoOBswqubwtsDfQERgCXSOoUEeeQtMomRsSWEXF1QxWRtAVwEfCFiNgK2Iekq7Fuvs7AHWneLsBvgTskdSnIdgxwPNAN2BQ4vYGityX5GfQkCahXAt8A9gD2A86WtGOadx1wKtCV5Gd3EPA9gIjYP82za/q+Ewue35mkNTmysOCI+DdwJnCDpM2Ba4HrImJaA/U12yg4QFlzdQFeb6QL7ljgvIhYGhGvAecC3yy4via9viYi/g68DexcYn3eAwZJ6hARiyNidj15DgXmRsSfImJtREwAngMOL8hzbUS8EBGrgEkkwXVD1pCMt60BbiIJPhdGxMq0/NnAJwAi4omIeCQt9yXgCuCzTXincyJidVqfD4iIK4G5wKNAD5L/ITDb6DlAWXMtA7o2MjayHfByweeX07T1z6gT4P4DbFlsRSLiHeDrwInAYkl3SBrQhPrU1qlnwedXi6jPsohYl57XBpAlBddX1d4vqb+k2yW9KuktkhZivd2HBV6LiP82kudKYBBwcUSsbiSv2UbBAcqa62Hgv8ARDeRZRNI9VatPmlaKd4DNCz5vW3gxIu6KiINJWhLPkfzibqw+tXV6pcQ6FeMyknr1i4iOwE8ANXJPg1NtJW1JMknlauBnaRem2UbPAcqaJSLeJBl3uSSdHLC5pE0kfUHSr9NsE4CzJG2TTjY4G7h+Q89sxJPA/pL6pBM0xtRekNRd0pfSsajVJF2F6+p5xt+B/pKOkdRO0teBgcDtJdapGFsBbwFvp627k+pcXwLs+KG7GnYh8EREnEAytnZ5s2tplgEOUNZsEfFbku9AnQW8BiwEvg/8Nc3yv8DjwNPALGBmmlZKWfcAE9NnPcEHg0oV8COSFtJykrGd79XzjGXAYWneZcAZwGER8XopdSrS6SQTMFaStO4m1rn+M2B8Osvva409TNJQYAhJtyYkfw+7185eNNuY+Yu6ZmaWSW5BmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAmZlZJjlAWcVIWifpSUnPSLpZ0ubNeNZ1kr6anl8laWADeQ+QtE8JZbwkqWtT0+vkebvIsn4m6fRi62iWJw5QVkmrIuKTETEIeBc4sfCipOpSHhoRJ0TEsw1kOQAoOkCZWetygLKsuB/4aNq6uVfSjcAsSdWSLpD0mKSnJX0XQIk/SHpW0h1At9oHSZomac/0fIikmZKekjRV0g4kgfDUtPW2n6RtJN2SlvGYpH3Te7tIulvSvyRdAaixl5D0V0lPSJotaWSda/+X1mWqpG3StJ0kTUnvuV/SgBb5aZrlQLtKV8BMUjvgC8CUNGkwMCgi5qe/5N+MiE9Jag88KOluYDdgZ+DjQHfgWeCaOs/dBrgS2D99VueIWC7pcuDtiPhNmu9G4HcR8YCkPsBdwMeAc4AHIuI8SYcCHwg4G/DttIwOwGOSbomIZcAWwMyI+JGks9Nnfx8YB5wYEXMlfRq4FDiwhB+jWe44QFkldZD0ZHp+P3A1SdfbjIiYn6Z/HvhE7fgSsDXQD9gfmBAR64BFkv5Zz/P3AqbXPisilm+gHp8DBkrrG0gdJW2VlvHl9N47JK1owjuNknRket47resy4D1gYpp+PfAXSVum73tzQdntm1CGWZvgAGWVtCoiPlmYkP6ifqcwCTglIu6qk++LQDTyfDUhDyRd3XtHxKp66tKU+2vzH0AS7PaOiP9ImgZstoHskZb7Rt2fgZklPAZlWXcXcJKkTQAk9Ze0BTAdGJaOUfUA/qeeex8GPiupb3pv5zR9JbBVQb67SbrbSPN9Mj2dDhybpn0B6NRIXbcGVqTBaQBJC65WFVDbCjyGpOvwLWC+pKPSMiRp10bKMGszHKAs664iGV+aKekZ4AqSlv+twFxgFnAZcF/dGyPiNZJxo79Ieor3u9j+BhxZO0kCGAXsmU7CeJb3ZxOeC+wvaSZJV+OCRuo6BWgn6Wng58AjBdfeAXaR9ATJGNN5afqxwIi0frOBoU34mZi1CYpocg+GmZlZq3ELyszMMskByszMMimzs/g69DnafY/W6lYtOLfSVbA2p3+jXwAvRrG/O1ctmNCi5bekzAYoMzMrnpSfjjEHKDOzHFGORm4coMzMcsQtKDMzyyQHKDMzy6SCdR03eg5QZma54haUmZllUJ66+PLzJmZmhlRV1NH483SNpKXpWph1r50uKSR1LUgbI2mepOclHVKQvoekWem1i9SEvkgHKDOzHBFVRR1NcB0w5EPlSL2BgylYRFnSQGAYsEt6z6WSqtPLl5Es3twvPT70zLocoMzMcqSlW1ARMR2ob7PP3wFn8ME904YCN0XE6nSj0HnA4HRLnI4R8XAkK5T/ETiisbIdoMzMcqTYACVppKTHC46RjZehLwGvRMRTdS71BBYWfK5J03qm53XTG+RJEmZmOVLsJImIGAeMa/rztTnwU5I90j50ub4iGkhvkAOUmVmOqN5Y0KJ2AvoCT6XzHHqRbCg6mKRl1Lsgby9gUZreq570BrmLz8wsR1p6DKquiJgVEd0iYoeI2IEk+OweEa8Ck4FhktpL6ksyGWJGRCwGVkraK529dxxwW2NlOUCZmeVIGaaZTwAeBnaWVCNpxIbyRsRsYBLwLDAFODki1qWXTwKuIpk48W/gzsbKdhefmVmOtPQXdSPi6Eau71Dn81hgbD35HgcGFVO2A5SZWa7kp2PMAcrMLEfytNSRA5SZWY44QJmZWSZ5R10zM8skt6DMzCyTvGGhmZllkltQZmaWSR6DMjOzTHILyszMMskByszMMsldfGZmlk1uQZmZWRa5i8/MzDLJ34MyM7NM8hiUmZllkrv4zMwsm9zFZ2ZmmVTtAGVmZlnkFpSZmWVSfoagHKDMzPIk3IIyM7NMyk98coAyM8uVqvxEKAcoM7M8cRefmZllUn7iU57me5iZGVUq7miEpGskLZX0TEHaBZKek/S0pFslfaTg2hhJ8yQ9L+mQgvQ9JM1Kr12kJiwa6ABlZpYnUnFH464DhtRJuwcYFBGfAF4AxiRFayAwDNglvedSSdXpPZcBI4F+6VH3mR/iAGVmlicq8mhEREwHltdJuzsi1qYfHwF6pedDgZsiYnVEzAfmAYMl9QA6RsTDERHAH4EjGivbY1BmZnnS+rP4vg1MTM97kgSsWjVp2pr0vG56g9yCMjPLkyJbUJJGSnq84BjZ5KKknwJrgRsKSq8rGkhvkFtQZmY5UuxKEhExDhhXbDmShgOHAQel3XaQtIx6F2TrBSxK03vVk94gt6DMzPKkhWfx1UfSEOBM4EsR8Z+CS5OBYZLaS+pLMhliRkQsBlZK2iudvXcccFtj5bgFZWaWJy08BCVpAnAA0FVSDXAOyay99sA96WzxRyLixIiYLWkS8CxJ19/JEbEufdRJJDMCOwB3pkeDHKDMzPKkhVeSiIij60m+uoH8Y4Gx9aQ/DgwqpmwHKDOzPPFafGZmlkn5iU8OUGZmueLFYs3MLJMcoMzMLJNy9OUhBygzszxxC8rMzDIpP/EpT43B/Lr8gu/y8szLefyeX3/o2g9HHsqqBRPo0mmr9WmDBvRh2q3n8sQ/LuCxu39F+/abAHDXxP/HU/f+H4/c+UseufOXbNOlY6u9g+XX9OlPcMghJ3LwwSMZN+7mSlenzYsqFXVkmVtQG4E/3Xwfl4+/i6t+970PpPfq0ZkD9/s4C2peW59WXV3FNReezIgfXsKsOQvo/JEtWbNm7frrx//gEmY+/WKr1d3ybd26dZx33uVce+3P6d69C1/96mkceOCn+ehH+1S6am1Xjrr43ILaCDw44zmWv/H2h9J/fc5x/PQXNxIFawJ/bv9P8MycBcyaswCA5W+8zXvvNbposFlJnn56Lttv34Pevbdl00034dBD92fq1EcrXa22rYX3g6okB6iN1KEH78GiV5evD0S1+u3YgyCY/KfRPHTHLzjtxMM/cP2K33yXR+78JaNHHdma1bWcWrJkGdtu23X95+7du7BkybIK1shaY7HY1lK2Lj5JA0h2V+xJsu/HImByRMwpV5ltRYfNNuXM7x/BYd/4xYeutauuYp89d+Yzh5/Ff1at5s4JP2XmrBeZ9uBsjh/1BxYtWcGWW2zGhCtO5Ziv7MeNt9xfgTewvIj4cOtcOepi2ijl6OdflhaUpDOBm0gakDOAx9LzCZJGN3Df+o2z1r49rxxVy4Udt+/O9r23YcaUX/HcgxfRs0dnHv77L+i+zda8sng59z86h2UrVrLqv+8y5d4n2W1QXwAWLVkBwNvv/JeJf32QT+26UyVfw3Jg22278uqrr6//vGTJMrp161zBGpm7+Bo3AvhURJwfEdenx/nA4PRavSJiXETsGRF7ttvyo2Wq2sZv9vML2X73Exmw7ygG7DuKVxYvZ+8v/oQlr73JPdOfZtCAPnTYbFOqq6vYb6+PMWfuK1RXV62f6deuXTVf/NzuzH6hppGSzBr28Y/346WXFrFw4au8++4a7rhjOgceOLjS1Wrb3MXXqPeA7YCX66T3SK9ZEcZffAr77f0xunbainmP/oGf//bPjJ84rd68b7z5Dhdd9XceuH0sEcFd9z7JlH/+i807tGfy9aPZpF07qquruPeBWVxz49TWfRHLnXbtqjn77BM54YRzWLfuPb7ylc/Rr9/2la5W25bxoFMM1deH3OyHJrst/gGYCyxMk/sAHwW+HxFTGntGhz5He+qZtbpVC86tdBWszenfohFlxxNuLup354tXHZXZiFaWFlRETJHUn6RLrydJT2cN8FjB7opmZtbSctSCKtssvoh4D3ikXM83M7N65GgWn1eSMDPLE7egzMwsk3K0/IIDlJlZnriLz8zMsiiq89OEcoAyM8uT/MQnBygzs1zxJAkzM8ukHI1B5agxaGZmLb0Wn6RrJC2V9ExBWmdJ90iam/7ZqeDaGEnzJD0v6ZCC9D0kzUqvXaQmLHvvAGVmlictv5r5dcCQOmmjgakR0Q+Ymn5G0kBgGLBLes+lkqrTey4DRgL90qPuMz/EAcrMLEeiSkUdjT4vYjqwvE7yUGB8ej4eOKIg/aaIWB0R84F5wGBJPYCOEfFwJAvA/rHgng3yGJSZWZ60ziSJ7hGxGCAiFkvqlqb35INL3NWkaWvS87rpDXILyswsT6SijsKNYtNjZHNKryctGkhvkFtQZmZ5UmSzIyLGAeOKLGWJpB5p66kHsDRNrwF6F+TrBSxK03vVk94gt6DMzPKkyBZUiSYDw9Pz4cBtBenDJLWX1JdkMsSMtDtwpaS90tl7xxXcs0FuQZmZ5UkLj0FJmgAcAHSVVAOcA5wPTJI0AlgAHAUQEbMlTQKeBdYCJxfsAXgSyYzADsCd6dEgBygzszxp4QAVEUdv4NJBG8g/FhhbT/rjwKBiynaAMjPLkcjRShIOUGZmeZKjmQUOUGZmeeIWlJmZZZJXMzczs0xygDIzs0zKT3xygDIzy5OmLAC7sXCAMjPLE0+SMDOzTHILyszMMik/8ckByswsT6r8RV0zM8uiHA1BOUCZmeVJmwhQklby/o6Hta9cuzNiRETHMtfNzMyKpBxFqA0GqIjYqjUrYmZmzZej+NS0dW8lfUbS8el513SnRDMzy5jW2VC3dTQ6BiXpHGBPYGfgWmBT4Hpg3/JWzczMiqU2NovvSGA3YCZARCyS5O4/M7MMynqrqBhNCVDvRkRICgBJW5S5TmZmVqIcLSTRpDGoSZKuAD4i6TvAP4Ary1stMzMrRZsag4qI30g6GHgL6A+cHRH3lL1mZmZWtKwHnWI09Yu6s4AOJN+DmlW+6piZWXPk6XtQjXbxSToBmAF8Gfgq8Iikb5e7YmZmVjxVFXdkWVNaUD8GdouIZQCSugAPAdeUs2JmZla8HDWgmhSgaoCVBZ9XAgvLUx0zM2uONhGgJJ2Wnr4CPCrpNpIxqKEkXX5mZpYx1WXotpN0KnAC789DOB7YHJgI7AC8BHwtIlak+ccAI4B1wKiIuKuUcht6la3S49/AX3l/4djbgMWlFGZmZuXV0tPMJfUERgF7RsQgoBoYBowGpkZEP2Bq+hlJA9PruwBDgEslVZfyLg0tFntuKQ80M7PKKVMXXzugg6Q1JC2nRcAY4ID0+nhgGnAmSS/bTRGxGpgvaR4wGHi4lEIbJGkb4AySaLhZbXpEHFhsYWZmVl5q4aUkIuIVSb8BFgCrgLsj4m5J3SNicZpnsaRu6S09gUcKHlGTphWtKb2VNwDPAX2Bc0n6Gh8rpTAzMyuvYrv4JI2U9HjBMfKDz1MnklZRX2A7YAtJ32ioCvWkRT1pjWrKLL4uEXG1pB9ExH3AfZLuK6UwMzMrr2K7+CJiHDCugSyfA+ZHxGvJ8/UXYB9giaQeaeupB7A0zV8D9C64vxdJl2DRmtKCWpP+uVjSoZJ2Sws0M7OMKcNafAuAvSRtrmSZioOAOcBkYHiaZzjJBDrS9GGS2qd7B/ajxJnfTWlB/a+krYEfARcDHYFTSynMzMzKq6VXM4+IRyX9mWTLpbXAv0haXFuSLCY+giSIHZXmny1pEvBsmv/kiFhXStmKKKlrsOw69Dk6mxWzXFu1wJNXrbX1b9GQsu+tDxT1u/PBIz+T2a/2NvRF3YtpYGArIkaVpUZmZlayrK+vV4yGuvgeb7VamJlZi2gTSx1FxPjWrIiZmTVfnrbbaOp+UGZmthHIUXxygDIzyxMHKDMzy6Q2EaAqPYvvnZfPKufjzcxyqaW/B1VJnsVnZpYjbSJAeRafmdnGp0r5WeOgqdttnAkMxNttmJllWp5aUE3dbmMO3m7DzCzzqoo8sqwp9esSEVcDayLivoj4NrBXmetlZmYlqFIUdWRZU6aZf2C7DZJ9PbzdhplZBuWpi8/bbZiZ5UjWu+2K0WiAiojb09M3gf8pb3XMzKw52lQLStK11POF3XQsyszMMkQZH1cqRlO6+G4vON8MOJIS95c3M7PyalMtqIi4pfCzpAnAP8pWIzMzK1mbGoOqRz+gT0tXxMzMmi/rU8eL0ZQxqJV8cAzqVZKVJczMLGPaWhffVq1RETMza748dfE1+i6SpjYlzczMKq9KxR1Z1tB+UJsBmwNdJXUCal+lI7BdK9TNzMyK1FbGoL4L/JAkGD3B+wHqLeCS8lbLzMxKkfVWUTEa2g/qQuBCSadExMWtWCczMytRmxqDAt6T9JHaD5I6Sfpe+apkZmalKsdq5pI+IunPkp6TNEfS3pI6S7pH0tz0z04F+cdImifpeUmHlPwuTcjznYh4o/ZDRKwAvlNqgWZmVj5lmiRxITAlIgYAu5LsETgamBoR/YCp6WckDQSGAbsAQ4BLJVWX9C5NySNp/WukBW1aSmFmZlZeLR2gJHUE9geuBoiId9NGy1BgfJptPHBEej4UuCkiVkfEfGAeMLikd2lCnruASZIOknQgMAGYUkphZmZWXmXYUXdH4DXgWkn/knSVpC2A7hGxGCD9s1uavyewsOD+mjStaE1Z6uhMYCRwEslMvruBK0spzMzMyqtdVXHTzCWNJPkdX2tcRIwrfCSwO3BKRDwq6ULS7rwNPbKetJLmvjdlJYn3gMvTA0mfIdm48ORSCjQzs/IpdhZfGozGNZClBqiJiEfTz38mCVBLJPWIiMWSegBLC/L3Lri/FyXugNGkd5H0SUm/kvQS8HPguVIKMzOz8mrpMaiIeBVYKGnnNOkg4FlgMjA8TRsO3JaeTwaGSWovqS/JAuMzSnmXhlaS6E8yE+NoYBkwEVBEeFddM7OMKtOGhacAN0jaFHgROJ6kgTNJ0ghgAXAUQETMljSJJIitBU6OiHWlFNpQF99zwP3A4RExD0DSqaUUYmZmraMcK0lExJPAnvVcOmgD+ccCY5tbbkNdfF8h2VrjXklXSjqI+ge/zMwsI8owi69iNli/iLg1Ir4ODACmAacC3SVdJunzrVQ/MzMrQjlWkqiURgNoRLwTETdExGEkszGepOEphmZmViFtYruN+kTEcuCK9DAzs4zJetApRlEByszMsq2kRe8yygHKzCxHsj6uVAwHKDOzHHEXn5mZZZIDlJmZZVK1A5SZmWWRW1BmZpZJniRhZmaZ5BaUmZllkr8HZWZmmeQWlJmZZZLHoMzMLJM8zdzMzDLJXXxmZpZJDlBmZpZJDlBmZpZJ1Z4kYWZmWdToNukbEQcoM7MccRefmZllkgOUmZllksegzMwsk/LUgsrTeJqZWZtXpeKOppBULelfkm5PP3eWdI+kuemfnQryjpE0T9Lzkg5p1rs052YzM8uWcgQo4AfAnILPo4GpEdEPmJp+RtJAYBiwCzAEuFRSyQusO0CZmeVItYo7GiOpF3AocFVB8lBgfHo+HjiiIP2miFgdEfOBecDgUt/FY1BmZjlShtXMfw+cAWxVkNY9IhYDRMRiSd3S9J7AIwX5atK0krgFZWaWI1VFHpJGSnq84BhZ+yxJhwFLI+KJJhZfX5us5IjpFtRGbPHi1xl95kW8/voKVFXF1752MMcdd9j669dc/VcuuOCPPPTwdXTq1LGCNbW8GjPmQqZNe4wuXbbm9tsvqXR1jOJn8UXEOGDcBi7vC3xJ0heBzYCOkq4HlkjqkbaeegBL0/w1QO+C+3sBi4qr0fvcgtqIVVdXccaZw7nj7xcz8abzufGGO5k3byGQBK+HHnqaHtt1rXAtLc++/OWDuOqqn1W6GlagJcegImJMRPSKiB1IJj/8MyK+AUwGhqfZhgO3peeTgWGS2kvqC/QDZpT6Lg5QG7Fu3Tqzyy47AbDFlh3YaadeLFmyDIDzf3kNp//4m6jeFrdZy/jUpwax9dZbNZ7RWk2VoqijROcDB0uaCxycfiYiZgOTgGeBKcDJEbGu1ELcxZcTr9QsZc6c+ey6a3/++c8ZdO/ehQED+la6WmbWytqVqdkREdOAaen5MuCgDeQbC4xtiTJbvQUl6fgGrq0frBs37ubWrNZG7Z13VjFq1K8ZPebbVFdXc8Xlt3DKqGGVrpaZVUCxkySyrBItqHOBa+u7UDhY917Mzs+CUmW0Zs1afjDqAg4/fH8+//m9eOH5l6mpWcIRQ08DYMmSZXzly6czcdKv2GabTo08zcw2dspRr35ZApSkpzd0CehejjLboojgrLMuYcedevKt478EQP+dt+fBh65bn+egA7/Ln2+5wLP4zNqIHMWnsrWgugOHACvqpAt4qExltjkzZz7H5Nvuo3//7TnyiKTF9MNTj+Wzn92jwjWztuK00y5gxoxZrFjxFvvv/y1OOeUYjjrq85WuVpuWpxaUIlq+J03S1cC1EfFAPddujIhjGnuGu/isEqq0SaWrYG1O/xYNKTNfv6Oo3527dz00syGtLC2oiBjRwLVGg5OZmZVG3g/KzMyyKLPNoRI4QJmZ5UiexqAcoMzMciRH8ckByswsT/K05bsDlJlZjuQoPjlAmZnlicegzMwsk3IUnxygzMzyxAHKzMwyyZMkzMwsk3IUnxygzMzyxEsdmZlZJrkFZWZmmeRp5mZmlklZ38a9GA5QZmY54haUmZllUo7ikwOUmVmeuAVlZmaZlKP45ABlZpYneVpJIk8TPszM2jwVeTT6PKm3pHslzZE0W9IP0vTOku6RNDf9s1PBPWMkzZP0vKRDSn0XBygzsxyRoqijCdYCP4qIjwF7ASdLGgiMBqZGRD9gavqZ9NowYBdgCHCppOpS3sUByswsR1q6BRURiyNiZnq+EpgD9ASGAuPTbOOBI9LzocBNEbE6IuYD84DBpbyLA5SZWY5IxR3FPVs7ALsBjwLdI2IxJEEM6JZm6wksLLitJk0rmgOUmVmOFNuCkjRS0uMFx8h6nyttCdwC/DAi3mqkCnWVtIKtZ/GZmeVIsa2OiBgHjGsoj6RNSILTDRHxlzR5iaQeEbFYUg9gaZpeA/QuuL0XsKjIagFuQZmZ5UpLd/FJEnA1MCcifltwaTIwPD0fDtxWkD5MUntJfYF+wIxS3sUtKDOzXGnxL0LtC3wTmCXpyTTtJ8D5wCRJI4AFwFEAETFb0iTgWZIZgCdHxLpSClZENje3ei9mZ7NilmtV2qTSVbA2p3+LRpQVq28v6ndnp/aHZfarvW5BmZnliJSfkRsHKDOzXMlsg6hoDlBmZjmiHM19c4AyM8sRd/GZmVlGuYvPzMwySA5QZmaWRQ5QZmaWUR6DMjOzDFKxS5RnmAOUmVmuOECZmVkGeQzKzMwyymNQZmaWQW5BmZlZJnmShJmZZZQDlJmZZZAXizUzs4xyC8rMzDLIY1BmZpZRDlBmZpZBHoMyM7OMcgvKzMwyyF/UNTOzTPIkCTMzyyiPQZmZWQa5i8/MzDIqPwEqP21BMzNDUlFHE585RNLzkuZJGl3mV1jPAcrMLFeqijwaJqkauAT4AjAQOFrSwHLUvC4HKDOzHFGR/zXBYGBeRLwYEe8CNwFDy/oSqcyOQVVpl/x0pLYySSMjYlyl62Fth//NZUn/on53ShoJjCxIGlfn77InsLDgcw3w6dLr13RuQeXTyMazmLUo/5vbSEXEuIjYs+Co+z8a9QW8aI26OUCZmVlDaoDeBZ97AYtao2AHKDMza8hjQD9JfSVtCgwDJrdGwZkdg7Jm8ViAtTb/m8upiFgr6fvAXUA1cE1EzG6NshXRKl2JZmZmRXEXn5mZZZIDlJmZZZIDVI5UajkSa7skXSNpqaRnKl0Xyx8HqJyo5HIk1qZdBwypdCUsnxyg8qNiy5FY2xUR04Hlla6H5ZMDVH7UtxxJzwrVxcys2Ryg8qNiy5GYmZWDA1R+VGw5EjOzcnCAyo+KLUdiZlYODlA5ERFrgdrlSOYAk1prORJruyRNAB4GdpZUI2lEpetk+eGljszMLJPcgjIzs0xygDIzs0xygDIzs0xygDIzs0xygDIzs0xygDIzs0xygDIzs0z6/+0pUfiJZ6IEAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm = metrics.confusion_matrix(test_target ,np.round(predictions_array))\n",
    "class_names=[0,1] # name  of classes\n",
    "fig, ax = plt.subplots()\n",
    "tick_marks = np.arange(len(class_names))\n",
    "plt.xticks(tick_marks, class_names)\n",
    "plt.yticks(tick_marks, class_names)\n",
    "# create heatmap\n",
    "sns.heatmap(pd.DataFrame(cm), annot=True, cmap=\"YlGnBu\" ,fmt='g')\n",
    "ax.xaxis.set_label_position(\"top\")\n",
    "plt.tight_layout()\n",
    "plt.title('Confusion matrix', y=1.1)\n",
    "plt.ylabel('Actual label')\n",
    "plt.xlabel('Predicted label')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "3b088825",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.98      1.00      0.99      1465\n",
      "           1       1.00      0.04      0.08        25\n",
      "\n",
      "    accuracy                           0.98      1490\n",
      "   macro avg       0.99      0.52      0.53      1490\n",
      "weighted avg       0.98      0.98      0.98      1490\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(test_target , np.round(predictions_array)))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d27118c7",
   "metadata": {},
   "source": [
    "# Delete the endpoint and bucket"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "4a57f1ee",
   "metadata": {},
   "outputs": [],
   "source": [
    "##sess.delete_endpoint(endpoint_name=predictor.endpoint_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "159de4d7",
   "metadata": {},
   "outputs": [],
   "source": [
    "##bucket_to_delete = boto3.resource('s3').Bucket(bucket_name)\n",
    "##bucket_to_delete.objects.all().delete()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fde4c21c",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "conda_python3",
   "language": "python",
   "name": "conda_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": "3.8.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}