Fraud Detection Modeling on Amazon SageMaker - Advanced.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Fraud Detection Modeling on Amazon SageMaker - Advanced"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fraud Detection Using Machine Learning enables you to execute automated transaction processing on an example dataset or your own dataset. The included ML model detects potentially fraudulent activity and flags that activity for review."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## AWS Architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from IPython.display import Image\n",
    "file_name1 = './img/fraud-detection-using-machine-learning.png'\n",
    "Image(file_name1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np                                # For matrix operations and numerical processing\n",
    "import pandas as pd                               # For munging tabular data\n",
    "import matplotlib.pyplot as plt                   # For charts and visualizations\n",
    "%matplotlib inline\n",
    "from IPython.display import Image                 # For displaying images in the notebook\n",
    "from IPython.display import display               # For displaying outputs in the notebook\n",
    "from time import gmtime, strftime                 # For labeling SageMaker models, endpoints, etc.\n",
    "import sys                                        # For writing outputs to notebook\n",
    "import math                                       # For ceiling function\n",
    "import json                                       # For parsing hosting outputs\n",
    "import os                                         # For manipulating filepath names\n",
    "import jsonlines\n",
    "import boto3 \n",
    "import csv\n",
    "import io\n",
    "import re\n",
    "import s3fs\n",
    "import copy\n",
    "import random\n",
    "import time\n",
    "from time import strftime, gmtime\n",
    "from scipy.sparse import lil_matrix\n",
    "\n",
    "# plot aesthetics\n",
    "import seaborn as sns\n",
    "sns.set(color_codes=True)\n",
    "sns.set_context('paper')\n",
    "five_thirty_eight = [\"#30a2da\", \"#fc4f30\", \"#e5ae38\", \"#6d904f\", \"#8b8b8b\",]\n",
    "sns.set_palette(five_thirty_eight)\n",
    "\n",
    "%matplotlib inline\n",
    "\n",
    "from urllib.request import urlretrieve\n",
    "import zipfile\n",
    "from dateutil.parser import parse\n",
    "from random import shuffle\n",
    "import random\n",
    "import datetime\n",
    "import urllib\n",
    "\n",
    "import sagemaker                                 \n",
    "from sagemaker.predictor import csv_serializer \n",
    "from sagemaker.predictor import json_deserializer\n",
    "import sagemaker.amazon.common as smac\n",
    "from sagemaker import get_execution_role\n",
    "from sagemaker.tuner import HyperparameterTuner, ContinuousParameter\n",
    "from sagemaker.analytics import HyperparameterTuningJobAnalytics, TrainingJobAnalytics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sagemaker session : <sagemaker.session.Session object at 0x000002F569B2CD88>\n",
      "S3 bucket : sagemaker-us-west-2-570447867175\n",
      "Prefix : fraud-detec\n",
      "Region selected : us-west-2\n",
      "IAM role : arn:aws:iam::570447867175:role/SageMakerNotebookRole\n"
     ]
    }
   ],
   "source": [
    "sess   = sagemaker.Session()\n",
    "bucket = sess.default_bucket()                     \n",
    "prefix = 'fraud-detec'\n",
    "region = boto3.Session().region_name\n",
    "role = 'arn:aws:iam::570447867175:role/SageMakerNotebookRole' # pass your IAM role name\n",
    "\n",
    "print('Sagemaker session :', sess)\n",
    "print('S3 bucket :', bucket)\n",
    "print('Prefix :', prefix)\n",
    "print('Region selected :', region)\n",
    "print('IAM role :', role)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1. Load Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Data Acknowledgements** \n",
    "The dataset used to demonstrated the fraud detection solution has been collected and analysed during a research collaboration of Worldline and the Machine Learning Group (http://mlg.ulb.ac.be) of ULB (Université Libre de Bruxelles) on big data mining and fraud detection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# uncomment below code to download data from wget\n",
    "# %%bash\n",
    "# wget https://s3-us-west-2.amazonaws.com/sagemaker-e2e-solutions/fraud-detection/creditcardfraud.zip\n",
    "# unzip creditcardfraud.zip"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Time</th>\n",
       "      <th>V1</th>\n",
       "      <th>V2</th>\n",
       "      <th>V3</th>\n",
       "      <th>V4</th>\n",
       "      <th>V5</th>\n",
       "      <th>V6</th>\n",
       "      <th>V7</th>\n",
       "      <th>V8</th>\n",
       "      <th>V9</th>\n",
       "      <th>...</th>\n",
       "      <th>V21</th>\n",
       "      <th>V22</th>\n",
       "      <th>V23</th>\n",
       "      <th>V24</th>\n",
       "      <th>V25</th>\n",
       "      <th>V26</th>\n",
       "      <th>V27</th>\n",
       "      <th>V28</th>\n",
       "      <th>Amount</th>\n",
       "      <th>Class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>-1.359807</td>\n",
       "      <td>-0.072781</td>\n",
       "      <td>2.536347</td>\n",
       "      <td>1.378155</td>\n",
       "      <td>-0.338321</td>\n",
       "      <td>0.462388</td>\n",
       "      <td>0.239599</td>\n",
       "      <td>0.098698</td>\n",
       "      <td>0.363787</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.018307</td>\n",
       "      <td>0.277838</td>\n",
       "      <td>-0.110474</td>\n",
       "      <td>0.066928</td>\n",
       "      <td>0.128539</td>\n",
       "      <td>-0.189115</td>\n",
       "      <td>0.133558</td>\n",
       "      <td>-0.021053</td>\n",
       "      <td>149.62</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1.191857</td>\n",
       "      <td>0.266151</td>\n",
       "      <td>0.166480</td>\n",
       "      <td>0.448154</td>\n",
       "      <td>0.060018</td>\n",
       "      <td>-0.082361</td>\n",
       "      <td>-0.078803</td>\n",
       "      <td>0.085102</td>\n",
       "      <td>-0.255425</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.225775</td>\n",
       "      <td>-0.638672</td>\n",
       "      <td>0.101288</td>\n",
       "      <td>-0.339846</td>\n",
       "      <td>0.167170</td>\n",
       "      <td>0.125895</td>\n",
       "      <td>-0.008983</td>\n",
       "      <td>0.014724</td>\n",
       "      <td>2.69</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.0</td>\n",
       "      <td>-1.358354</td>\n",
       "      <td>-1.340163</td>\n",
       "      <td>1.773209</td>\n",
       "      <td>0.379780</td>\n",
       "      <td>-0.503198</td>\n",
       "      <td>1.800499</td>\n",
       "      <td>0.791461</td>\n",
       "      <td>0.247676</td>\n",
       "      <td>-1.514654</td>\n",
       "      <td>...</td>\n",
       "      <td>0.247998</td>\n",
       "      <td>0.771679</td>\n",
       "      <td>0.909412</td>\n",
       "      <td>-0.689281</td>\n",
       "      <td>-0.327642</td>\n",
       "      <td>-0.139097</td>\n",
       "      <td>-0.055353</td>\n",
       "      <td>-0.059752</td>\n",
       "      <td>378.66</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.0</td>\n",
       "      <td>-0.966272</td>\n",
       "      <td>-0.185226</td>\n",
       "      <td>1.792993</td>\n",
       "      <td>-0.863291</td>\n",
       "      <td>-0.010309</td>\n",
       "      <td>1.247203</td>\n",
       "      <td>0.237609</td>\n",
       "      <td>0.377436</td>\n",
       "      <td>-1.387024</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.108300</td>\n",
       "      <td>0.005274</td>\n",
       "      <td>-0.190321</td>\n",
       "      <td>-1.175575</td>\n",
       "      <td>0.647376</td>\n",
       "      <td>-0.221929</td>\n",
       "      <td>0.062723</td>\n",
       "      <td>0.061458</td>\n",
       "      <td>123.50</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2.0</td>\n",
       "      <td>-1.158233</td>\n",
       "      <td>0.877737</td>\n",
       "      <td>1.548718</td>\n",
       "      <td>0.403034</td>\n",
       "      <td>-0.407193</td>\n",
       "      <td>0.095921</td>\n",
       "      <td>0.592941</td>\n",
       "      <td>-0.270533</td>\n",
       "      <td>0.817739</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.009431</td>\n",
       "      <td>0.798278</td>\n",
       "      <td>-0.137458</td>\n",
       "      <td>0.141267</td>\n",
       "      <td>-0.206010</td>\n",
       "      <td>0.502292</td>\n",
       "      <td>0.219422</td>\n",
       "      <td>0.215153</td>\n",
       "      <td>69.99</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 31 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Time        V1        V2        V3        V4        V5        V6        V7  \\\n",
       "0   0.0 -1.359807 -0.072781  2.536347  1.378155 -0.338321  0.462388  0.239599   \n",
       "1   0.0  1.191857  0.266151  0.166480  0.448154  0.060018 -0.082361 -0.078803   \n",
       "2   1.0 -1.358354 -1.340163  1.773209  0.379780 -0.503198  1.800499  0.791461   \n",
       "3   1.0 -0.966272 -0.185226  1.792993 -0.863291 -0.010309  1.247203  0.237609   \n",
       "4   2.0 -1.158233  0.877737  1.548718  0.403034 -0.407193  0.095921  0.592941   \n",
       "\n",
       "         V8        V9  ...       V21       V22       V23       V24       V25  \\\n",
       "0  0.098698  0.363787  ... -0.018307  0.277838 -0.110474  0.066928  0.128539   \n",
       "1  0.085102 -0.255425  ... -0.225775 -0.638672  0.101288 -0.339846  0.167170   \n",
       "2  0.247676 -1.514654  ...  0.247998  0.771679  0.909412 -0.689281 -0.327642   \n",
       "3  0.377436 -1.387024  ... -0.108300  0.005274 -0.190321 -1.175575  0.647376   \n",
       "4 -0.270533  0.817739  ... -0.009431  0.798278 -0.137458  0.141267 -0.206010   \n",
       "\n",
       "        V26       V27       V28  Amount  Class  \n",
       "0 -0.189115  0.133558 -0.021053  149.62      0  \n",
       "1  0.125895 -0.008983  0.014724    2.69      0  \n",
       "2 -0.139097 -0.055353 -0.059752  378.66      0  \n",
       "3 -0.221929  0.062723  0.061458  123.50      0  \n",
       "4  0.502292  0.219422  0.215153   69.99      0  \n",
       "\n",
       "[5 rows x 31 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "path = 'C:\\\\Users\\\\bokhy\\\\Desktop\\\\Python\\\\github\\\\amazon-sagemaker\\\\data'  # Set Your PATH\n",
    "\n",
    "data = pd.read_csv(os.path.join(path, 'creditcard.csv'), delimiter=',')\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. Data Pre-processing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The class column corresponds to whether or not a transaction is fradulent. We see that the majority of data is non-fraudulent with only $492$ ($.173\\%$) of the data corresponding to fraudulent examples."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of frauds:  492\n",
      "Number of non-frauds:  284315\n",
      "Percentage of fradulent data: 0.1727485630620034\n"
     ]
    }
   ],
   "source": [
    "nonfrauds, frauds = data.groupby('Class').size()\n",
    "print('Number of frauds: ', frauds)\n",
    "print('Number of non-frauds: ', nonfrauds)\n",
    "print('Percentage of fradulent data:', 100.*frauds/(frauds + nonfrauds))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This dataset has 28 columns, $V_i$ for $i=1..28$ of anonymized features along with columns for time, amount, and class. We already know that the columns $V_i$ have been normalized to have $0$ mean and unit standard deviation as the result of a PCA."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_columns = data.columns[:-1]\n",
    "label_column = data.columns[-1]\n",
    "\n",
    "features = data[feature_columns].values.astype('float32')\n",
    "labels = (data[label_column].values).astype('float32')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3. Start Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Train/Test Split\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    features, labels, test_size=0.1, random_state=623)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### (1) Unsupervised Learning"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In a fraud detection scenario, commonly we will have very few labeled examples, and it's possible that labeling fraud takes a very long time. We would like then to extract information from the unlabeled data we have at hand as well. Anomaly detection is a form of unsupervised learning where we try to identify anomalous examples based solely on their feature characteristics. **Random Cut Forest** is a state-of-the-art anomaly detection algorithm that is both accurate and scalable. We will train such a model on our training data and evaluate its performance on our test set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sagemaker import RandomCutForest\n",
    "\n",
    "# specify general training job information\n",
    "rcf = RandomCutForest(role=role,\n",
    "                      train_instance_count=1,\n",
    "                      train_instance_type='ml.m4.xlarge',\n",
    "                      data_location='s3://{}/{}/'.format(bucket, prefix),\n",
    "                      output_path='s3://{}/{}/output'.format(bucket, prefix),\n",
    "                      num_samples_per_tree=512,\n",
    "                      num_trees=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "'get_image_uri' method will be deprecated in favor of 'ImageURIProvider' class in SageMaker Python SDK v2.\n",
      "'s3_input' class will be renamed to 'TrainingInput' in SageMaker Python SDK v2.\n",
      "'get_image_uri' method will be deprecated in favor of 'ImageURIProvider' class in SageMaker Python SDK v2.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-11-04 03:51:09 Starting - Starting the training job...\n",
      "2020-11-04 03:51:11 Starting - Launching requested ML instances......\n",
      "2020-11-04 03:52:20 Starting - Preparing the instances for training...\n",
      "2020-11-04 03:53:12 Downloading - Downloading input data...\n",
      "2020-11-04 03:53:25 Training - Downloading the training image...\n",
      "2020-11-04 03:54:15 Training - Training image download completed. Training in progress..\u001b[34mDocker entrypoint called with argument(s): train\u001b[0m\n",
      "\u001b[34mRunning default environment configuration script\u001b[0m\n",
      "\u001b[34m/opt/amazon/lib/python2.7/site-packages/scipy/_lib/_numpy_compat.py:10: DeprecationWarning: Importing from numpy.testing.nosetester is deprecated, import from numpy.testing instead.\n",
      "  from numpy.testing.nosetester import import_nose\u001b[0m\n",
      "\u001b[34m/opt/amazon/lib/python2.7/site-packages/scipy/stats/morestats.py:12: DeprecationWarning: Importing from numpy.testing.decorators is deprecated, import from numpy.testing instead.\n",
      "  from numpy.testing.decorators import setastest\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Reading default configuration from /opt/amazon/lib/python2.7/site-packages/algorithm/resources/default-conf.json: {u'_ftp_port': 8999, u'num_samples_per_tree': 256, u'_tuning_objective_metric': u'', u'_num_gpus': u'auto', u'_log_level': u'info', u'_kvstore': u'dist_async', u'force_dense': u'true', u'epochs': 1, u'num_trees': 100, u'eval_metrics': [u'accuracy', u'precision_recall_fscore'], u'_num_kv_servers': u'auto', u'mini_batch_size': 1000}\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Merging with provided configuration from /opt/ml/input/config/hyperparameters.json: {u'mini_batch_size': u'1000', u'feature_dim': u'30', u'num_samples_per_tree': u'512', u'num_trees': u'50'}\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Final configuration: {u'_ftp_port': 8999, u'num_samples_per_tree': u'512', u'_tuning_objective_metric': u'', u'_num_gpus': u'auto', u'_log_level': u'info', u'_kvstore': u'dist_async', u'force_dense': u'true', u'epochs': 1, u'feature_dim': u'30', u'num_trees': u'50', u'eval_metrics': [u'accuracy', u'precision_recall_fscore'], u'_num_kv_servers': u'auto', u'mini_batch_size': u'1000'}\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 WARNING 139864704436032] Loggers have already been setup.\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Launching parameter server for role scheduler\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] {'ECS_CONTAINER_METADATA_URI': 'http://169.254.170.2/v3/62503c69-887d-4c0a-afcd-347a211c8ec4', 'ECS_CONTAINER_METADATA_URI_V4': 'http://169.254.170.2/v4/62503c69-887d-4c0a-afcd-347a211c8ec4', 'PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION_VERSION': '2', 'PATH': '/opt/amazon/bin:/usr/local/nvidia/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/opt/amazon/bin:/opt/amazon/bin', 'SAGEMAKER_HTTP_PORT': '8080', 'HOME': '/root', 'PYTHONUNBUFFERED': 'TRUE', 'CANONICAL_ENVROOT': '/opt/amazon', 'LD_LIBRARY_PATH': '/opt/amazon/lib/python2.7/site-packages/cv2/../../../../lib:/usr/local/nvidia/lib64:/opt/amazon/lib', 'MXNET_KVSTORE_BIGARRAY_BOUND': '400000000', 'LANG': 'en_US.utf8', 'DMLC_INTERFACE': 'eth0', 'SHLVL': '1', 'AWS_REGION': 'us-west-2', 'SAGEMAKER_METRICS_DIRECTORY': '/opt/ml/output/metrics/sagemaker', 'NVIDIA_VISIBLE_DEVICES': 'void', 'TRAINING_JOB_NAME': 'randomcutforest-2020-11-04-03-51-09-065', 'PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION': 'cpp', 'ENVROOT': '/opt/amazon', 'SAGEMAKER_DATA_PATH': '/opt/ml', 'NVIDIA_DRIVER_CAPABILITIES': 'compute,utility', 'NVIDIA_REQUIRE_CUDA': 'cuda>=9.0', 'OMP_NUM_THREADS': '2', 'HOSTNAME': 'ip-10-0-123-196.us-west-2.compute.internal', 'AWS_CONTAINER_CREDENTIALS_RELATIVE_URI': '/v2/credentials/dc811a6b-de6a-4abd-bf77-4de0f2da04a9', 'PWD': '/', 'TRAINING_JOB_ARN': 'arn:aws:sagemaker:us-west-2:570447867175:training-job/randomcutforest-2020-11-04-03-51-09-065', 'AWS_EXECUTION_ENV': 'AWS_ECS_EC2'}\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] envs={'ECS_CONTAINER_METADATA_URI': 'http://169.254.170.2/v3/62503c69-887d-4c0a-afcd-347a211c8ec4', 'ECS_CONTAINER_METADATA_URI_V4': 'http://169.254.170.2/v4/62503c69-887d-4c0a-afcd-347a211c8ec4', 'PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION_VERSION': '2', 'DMLC_NUM_WORKER': '1', 'DMLC_PS_ROOT_PORT': '9000', 'PATH': '/opt/amazon/bin:/usr/local/nvidia/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/opt/amazon/bin:/opt/amazon/bin', 'SAGEMAKER_HTTP_PORT': '8080', 'HOME': '/root', 'PYTHONUNBUFFERED': 'TRUE', 'CANONICAL_ENVROOT': '/opt/amazon', 'LD_LIBRARY_PATH': '/opt/amazon/lib/python2.7/site-packages/cv2/../../../../lib:/usr/local/nvidia/lib64:/opt/amazon/lib', 'MXNET_KVSTORE_BIGARRAY_BOUND': '400000000', 'LANG': 'en_US.utf8', 'DMLC_INTERFACE': 'eth0', 'SHLVL': '1', 'DMLC_PS_ROOT_URI': '10.0.123.196', 'AWS_REGION': 'us-west-2', 'SAGEMAKER_METRICS_DIRECTORY': '/opt/ml/output/metrics/sagemaker', 'NVIDIA_VISIBLE_DEVICES': 'void', 'TRAINING_JOB_NAME': 'randomcutforest-2020-11-04-03-51-09-065', 'PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION': 'cpp', 'ENVROOT': '/opt/amazon', 'SAGEMAKER_DATA_PATH': '/opt/ml', 'NVIDIA_DRIVER_CAPABILITIES': 'compute,utility', 'NVIDIA_REQUIRE_CUDA': 'cuda>=9.0', 'OMP_NUM_THREADS': '2', 'HOSTNAME': 'ip-10-0-123-196.us-west-2.compute.internal', 'AWS_CONTAINER_CREDENTIALS_RELATIVE_URI': '/v2/credentials/dc811a6b-de6a-4abd-bf77-4de0f2da04a9', 'DMLC_ROLE': 'scheduler', 'PWD': '/', 'DMLC_NUM_SERVER': '1', 'TRAINING_JOB_ARN': 'arn:aws:sagemaker:us-west-2:570447867175:training-job/randomcutforest-2020-11-04-03-51-09-065', 'AWS_EXECUTION_ENV': 'AWS_ECS_EC2'}\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Launching parameter server for role server\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] {'ECS_CONTAINER_METADATA_URI': 'http://169.254.170.2/v3/62503c69-887d-4c0a-afcd-347a211c8ec4', 'ECS_CONTAINER_METADATA_URI_V4': 'http://169.254.170.2/v4/62503c69-887d-4c0a-afcd-347a211c8ec4', 'PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION_VERSION': '2', 'PATH': '/opt/amazon/bin:/usr/local/nvidia/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/opt/amazon/bin:/opt/amazon/bin', 'SAGEMAKER_HTTP_PORT': '8080', 'HOME': '/root', 'PYTHONUNBUFFERED': 'TRUE', 'CANONICAL_ENVROOT': '/opt/amazon', 'LD_LIBRARY_PATH': '/opt/amazon/lib/python2.7/site-packages/cv2/../../../../lib:/usr/local/nvidia/lib64:/opt/amazon/lib', 'MXNET_KVSTORE_BIGARRAY_BOUND': '400000000', 'LANG': 'en_US.utf8', 'DMLC_INTERFACE': 'eth0', 'SHLVL': '1', 'AWS_REGION': 'us-west-2', 'SAGEMAKER_METRICS_DIRECTORY': '/opt/ml/output/metrics/sagemaker', 'NVIDIA_VISIBLE_DEVICES': 'void', 'TRAINING_JOB_NAME': 'randomcutforest-2020-11-04-03-51-09-065', 'PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION': 'cpp', 'ENVROOT': '/opt/amazon', 'SAGEMAKER_DATA_PATH': '/opt/ml', 'NVIDIA_DRIVER_CAPABILITIES': 'compute,utility', 'NVIDIA_REQUIRE_CUDA': 'cuda>=9.0', 'OMP_NUM_THREADS': '2', 'HOSTNAME': 'ip-10-0-123-196.us-west-2.compute.internal', 'AWS_CONTAINER_CREDENTIALS_RELATIVE_URI': '/v2/credentials/dc811a6b-de6a-4abd-bf77-4de0f2da04a9', 'PWD': '/', 'TRAINING_JOB_ARN': 'arn:aws:sagemaker:us-west-2:570447867175:training-job/randomcutforest-2020-11-04-03-51-09-065', 'AWS_EXECUTION_ENV': 'AWS_ECS_EC2'}\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] envs={'ECS_CONTAINER_METADATA_URI': 'http://169.254.170.2/v3/62503c69-887d-4c0a-afcd-347a211c8ec4', 'ECS_CONTAINER_METADATA_URI_V4': 'http://169.254.170.2/v4/62503c69-887d-4c0a-afcd-347a211c8ec4', 'PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION_VERSION': '2', 'DMLC_NUM_WORKER': '1', 'DMLC_PS_ROOT_PORT': '9000', 'PATH': '/opt/amazon/bin:/usr/local/nvidia/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/opt/amazon/bin:/opt/amazon/bin', 'SAGEMAKER_HTTP_PORT': '8080', 'HOME': '/root', 'PYTHONUNBUFFERED': 'TRUE', 'CANONICAL_ENVROOT': '/opt/amazon', 'LD_LIBRARY_PATH': '/opt/amazon/lib/python2.7/site-packages/cv2/../../../../lib:/usr/local/nvidia/lib64:/opt/amazon/lib', 'MXNET_KVSTORE_BIGARRAY_BOUND': '400000000', 'LANG': 'en_US.utf8', 'DMLC_INTERFACE': 'eth0', 'SHLVL': '1', 'DMLC_PS_ROOT_URI': '10.0.123.196', 'AWS_REGION': 'us-west-2', 'SAGEMAKER_METRICS_DIRECTORY': '/opt/ml/output/metrics/sagemaker', 'NVIDIA_VISIBLE_DEVICES': 'void', 'TRAINING_JOB_NAME': 'randomcutforest-2020-11-04-03-51-09-065', 'PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION': 'cpp', 'ENVROOT': '/opt/amazon', 'SAGEMAKER_DATA_PATH': '/opt/ml', 'NVIDIA_DRIVER_CAPABILITIES': 'compute,utility', 'NVIDIA_REQUIRE_CUDA': 'cuda>=9.0', 'OMP_NUM_THREADS': '2', 'HOSTNAME': 'ip-10-0-123-196.us-west-2.compute.internal', 'AWS_CONTAINER_CREDENTIALS_RELATIVE_URI': '/v2/credentials/dc811a6b-de6a-4abd-bf77-4de0f2da04a9', 'DMLC_ROLE': 'server', 'PWD': '/', 'DMLC_NUM_SERVER': '1', 'TRAINING_JOB_ARN': 'arn:aws:sagemaker:us-west-2:570447867175:training-job/randomcutforest-2020-11-04-03-51-09-065', 'AWS_EXECUTION_ENV': 'AWS_ECS_EC2'}\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Environment: {'ECS_CONTAINER_METADATA_URI': 'http://169.254.170.2/v3/62503c69-887d-4c0a-afcd-347a211c8ec4', 'ECS_CONTAINER_METADATA_URI_V4': 'http://169.254.170.2/v4/62503c69-887d-4c0a-afcd-347a211c8ec4', 'PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION_VERSION': '2', 'DMLC_PS_ROOT_PORT': '9000', 'DMLC_NUM_WORKER': '1', 'SAGEMAKER_HTTP_PORT': '8080', 'PATH': '/opt/amazon/bin:/usr/local/nvidia/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/opt/amazon/bin:/opt/amazon/bin', 'PYTHONUNBUFFERED': 'TRUE', 'CANONICAL_ENVROOT': '/opt/amazon', 'LD_LIBRARY_PATH': '/opt/amazon/lib/python2.7/site-packages/cv2/../../../../lib:/usr/local/nvidia/lib64:/opt/amazon/lib', 'MXNET_KVSTORE_BIGARRAY_BOUND': '400000000', 'LANG': 'en_US.utf8', 'DMLC_INTERFACE': 'eth0', 'SHLVL': '1', 'DMLC_PS_ROOT_URI': '10.0.123.196', 'AWS_REGION': 'us-west-2', 'SAGEMAKER_METRICS_DIRECTORY': '/opt/ml/output/metrics/sagemaker', 'NVIDIA_VISIBLE_DEVICES': 'void', 'TRAINING_JOB_NAME': 'randomcutforest-2020-11-04-03-51-09-065', 'HOME': '/root', 'PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION': 'cpp', 'ENVROOT': '/opt/amazon', 'SAGEMAKER_DATA_PATH': '/opt/ml', 'NVIDIA_DRIVER_CAPABILITIES': 'compute,utility', 'NVIDIA_REQUIRE_CUDA': 'cuda>=9.0', 'OMP_NUM_THREADS': '2', 'HOSTNAME': 'ip-10-0-123-196.us-west-2.compute.internal', 'AWS_CONTAINER_CREDENTIALS_RELATIVE_URI': '/v2/credentials/dc811a6b-de6a-4abd-bf77-4de0f2da04a9', 'DMLC_ROLE': 'worker', 'PWD': '/', 'DMLC_NUM_SERVER': '1', 'TRAINING_JOB_ARN': 'arn:aws:sagemaker:us-west-2:570447867175:training-job/randomcutforest-2020-11-04-03-51-09-065', 'AWS_EXECUTION_ENV': 'AWS_ECS_EC2'}\u001b[0m\n",
      "\u001b[34mProcess 33 is a shell:scheduler.\u001b[0m\n",
      "\u001b[34mProcess 34 is a shell:server.\u001b[0m\n",
      "\u001b[34mProcess 1 is a worker.\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Using default worker.\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Loaded iterator creator application/x-recordio-protobuf for content type ('application/x-recordio-protobuf', '1.0')\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Checkpoint loading and saving are disabled.\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Verifying hyperparamemters...\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Hyperparameters are correct.\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Validating that feature_dim agrees with dimensions in training data...\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] feature_dim is correct.\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Validating memory limits...\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Available memory in bytes: 15256342528\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Estimated sample size in bytes: 6144000\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Estimated memory needed to build the forest in bytes: 30720000\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Memory limits validated.\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Starting cluster sharing facilities...\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139864704436032] Create Store: dist_async\u001b[0m\n",
      "\u001b[34m[I 20-11-04 03:54:20] >>> starting FTP server on 0.0.0.0:8999, pid=1 <<<\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139863236003584] >>> starting FTP server on 0.0.0.0:8999, pid=1 <<<\u001b[0m\n",
      "\u001b[34m[I 20-11-04 03:54:20] poller: <class 'pyftpdlib.ioloop.Epoll'>\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139863236003584] poller: <class 'pyftpdlib.ioloop.Epoll'>\u001b[0m\n",
      "\u001b[34m[I 20-11-04 03:54:20] masquerade (NAT) address: None\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139863236003584] masquerade (NAT) address: None\u001b[0m\n",
      "\u001b[34m[I 20-11-04 03:54:20] passive ports: None\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139863236003584] passive ports: None\u001b[0m\n",
      "\u001b[34m[I 20-11-04 03:54:20] use sendfile(2): False\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:20 INFO 139863236003584] use sendfile(2): False\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] Cluster sharing facilities started.\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] Verifying all workers are accessible...\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] All workers accessible.\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] Initializing Sampler...\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] Sampler correctly initialized.\u001b[0m\n",
      "\u001b[34m#metrics {\"Metrics\": {\"initialize.time\": {\"count\": 1, \"max\": 815.1669502258301, \"sum\": 815.1669502258301, \"min\": 815.1669502258301}}, \"EndTime\": 1604462061.078537, \"Dimensions\": {\"Host\": \"algo-1\", \"Operation\": \"training\", \"Algorithm\": \"RandomCutForest\"}, \"StartTime\": 1604462060.254246}\n",
      "\u001b[0m\n",
      "\u001b[34m#metrics {\"Metrics\": {\"Max Batches Seen Between Resets\": {\"count\": 1, \"max\": 0, \"sum\": 0.0, \"min\": 0}, \"Number of Batches Since Last Reset\": {\"count\": 1, \"max\": 0, \"sum\": 0.0, \"min\": 0}, \"Number of Records Since Last Reset\": {\"count\": 1, \"max\": 0, \"sum\": 0.0, \"min\": 0}, \"Total Batches Seen\": {\"count\": 1, \"max\": 0, \"sum\": 0.0, \"min\": 0}, \"Total Records Seen\": {\"count\": 1, \"max\": 0, \"sum\": 0.0, \"min\": 0}, \"Max Records Seen Between Resets\": {\"count\": 1, \"max\": 0, \"sum\": 0.0, \"min\": 0}, \"Reset Count\": {\"count\": 1, \"max\": 0, \"sum\": 0.0, \"min\": 0}}, \"EndTime\": 1604462061.078801, \"Dimensions\": {\"Host\": \"algo-1\", \"Meta\": \"init_train_data_iter\", \"Operation\": \"training\", \"Algorithm\": \"RandomCutForest\"}, \"StartTime\": 1604462061.07874}\n",
      "\u001b[0m\n",
      "\u001b[34m[2020-11-04 03:54:21.079] [tensorio] [info] epoch_stats={\"data_pipeline\": \"/opt/ml/input/data/train\", \"epoch\": 0, \"duration\": 824, \"num_examples\": 1, \"num_bytes\": 148000}\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] Sampling training data...\u001b[0m\n",
      "\u001b[34m[2020-11-04 03:54:21.624] [tensorio] [info] epoch_stats={\"data_pipeline\": \"/opt/ml/input/data/train\", \"epoch\": 1, \"duration\": 545, \"num_examples\": 257, \"num_bytes\": 37936248}\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] Sampling training data completed.\u001b[0m\n",
      "\u001b[34m#metrics {\"Metrics\": {\"epochs\": {\"count\": 1, \"max\": 1, \"sum\": 1.0, \"min\": 1}, \"update.time\": {\"count\": 1, \"max\": 563.0300045013428, \"sum\": 563.0300045013428, \"min\": 563.0300045013428}}, \"EndTime\": 1604462061.642336, \"Dimensions\": {\"Host\": \"algo-1\", \"Operation\": \"training\", \"Algorithm\": \"RandomCutForest\"}, \"StartTime\": 1604462061.078666}\n",
      "\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] Early stop condition met. Stopping training.\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] #progress_metric: host=algo-1, completed 100 % epochs\u001b[0m\n",
      "\u001b[34m#metrics {\"Metrics\": {\"Max Batches Seen Between Resets\": {\"count\": 1, \"max\": 257, \"sum\": 257.0, \"min\": 257}, \"Number of Batches Since Last Reset\": {\"count\": 1, \"max\": 257, \"sum\": 257.0, \"min\": 257}, \"Number of Records Since Last Reset\": {\"count\": 1, \"max\": 256326, \"sum\": 256326.0, \"min\": 256326}, \"Total Batches Seen\": {\"count\": 1, \"max\": 257, \"sum\": 257.0, \"min\": 257}, \"Total Records Seen\": {\"count\": 1, \"max\": 256326, \"sum\": 256326.0, \"min\": 256326}, \"Max Records Seen Between Resets\": {\"count\": 1, \"max\": 256326, \"sum\": 256326.0, \"min\": 256326}, \"Reset Count\": {\"count\": 1, \"max\": 1, \"sum\": 1.0, \"min\": 1}}, \"EndTime\": 1604462061.64275, \"Dimensions\": {\"Host\": \"algo-1\", \"Meta\": \"training_data_iter\", \"Operation\": \"training\", \"Algorithm\": \"RandomCutForest\", \"epoch\": 0}, \"StartTime\": 1604462061.079258}\n",
      "\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] #throughput_metric: host=algo-1, train throughput=454772.176819 records/second\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] Master node: building Random Cut Forest...\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] Gathering samples...\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] 25600 samples gathered\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] Building Random Cut Forest...\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:21 INFO 139864704436032] Random Cut Forest built: \n",
      "\u001b[0m\n",
      "\u001b[34mForestInfo{num_trees: 50, num_samples_in_forest: 25600, num_samples_per_tree: 512, sample_dim: 30, shingle_size: 1, trees_num_nodes: [953, 919, 943, 955, 961, 907, 941, 929, 923, 913, 935, 921, 933, 939, 931, 943, 919, 941, 943, 929, 945, 935, 941, 915, 925, 917, 925, 933, 921, 917, 941, 925, 923, 929, 925, 937, 925, 939, 945, 921, 939, 923, 925, 921, 933, 915, 929, 945, 921, 929, ], trees_depth: [23, 21, 20, 23, 22, 20, 20, 23, 20, 18, 21, 18, 20, 22, 21, 22, 19, 23, 22, 21, 23, 21, 21, 20, 19, 20, 27, 25, 23, 20, 23, 19, 18, 21, 21, 23, 21, 20, 23, 23, 28, 21, 24, 20, 20, 19, 20, 21, 19, 17, ], max_num_nodes: 961, min_num_nodes: 907, avg_num_nodes: 930, max_tree_depth: 28, min_tree_depth: 17, avg_tree_depth: 21, mem_size: 26436304}\u001b[0m\n",
      "\u001b[34m#metrics {\"Metrics\": {\"finalize.time\": {\"count\": 1, \"max\": 108.92200469970703, \"sum\": 108.92200469970703, \"min\": 108.92200469970703}, \"model.bytes\": {\"count\": 1, \"max\": 26436304, \"sum\": 26436304.0, \"min\": 26436304}, \"fit_model.time\": {\"count\": 1, \"max\": 73.64797592163086, \"sum\": 73.64797592163086, \"min\": 73.64797592163086}}, \"EndTime\": 1604462061.75199, \"Dimensions\": {\"Host\": \"algo-1\", \"Operation\": \"training\", \"Algorithm\": \"RandomCutForest\"}, \"StartTime\": 1604462061.642455}\n",
      "\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:22 INFO 139864704436032] Master node: Serializing the RandomCutForest model\u001b[0m\n",
      "\u001b[34m#metrics {\"Metrics\": {\"serialize_model.time\": {\"count\": 1, \"max\": 715.1448726654053, \"sum\": 715.1448726654053, \"min\": 715.1448726654053}}, \"EndTime\": 1604462062.467314, \"Dimensions\": {\"Host\": \"algo-1\", \"Operation\": \"training\", \"Algorithm\": \"RandomCutForest\"}, \"StartTime\": 1604462061.752095}\n",
      "\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:22 INFO 139864704436032] Test data is not provided.\u001b[0m\n",
      "\u001b[34m[I 20-11-04 03:54:22] >>> shutting down FTP server (0 active fds) <<<\u001b[0m\n",
      "\u001b[34m[11/04/2020 03:54:22 INFO 139863236003584] >>> shutting down FTP server (0 active fds) <<<\u001b[0m\n",
      "\u001b[34m#metrics {\"Metrics\": {\"totaltime\": {\"count\": 1, \"max\": 2365.2350902557373, \"sum\": 2365.2350902557373, \"min\": 2365.2350902557373}, \"setuptime\": {\"count\": 1, \"max\": 24.873018264770508, \"sum\": 24.873018264770508, \"min\": 24.873018264770508}}, \"EndTime\": 1604462062.584308, \"Dimensions\": {\"Host\": \"algo-1\", \"Operation\": \"training\", \"Algorithm\": \"RandomCutForest\"}, \"StartTime\": 1604462062.467434}\n",
      "\u001b[0m\n",
      "\n",
      "2020-11-04 03:54:43 Uploading - Uploading generated training model\n",
      "2020-11-04 03:54:43 Completed - Training job completed\n",
      "Training seconds: 91\n",
      "Billable seconds: 91\n"
     ]
    }
   ],
   "source": [
    "rcf.fit(rcf.record_set(X_train))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Parameter image will be renamed to image_uri in SageMaker Python SDK v2.\n",
      "Using already existing model: randomcutforest-2020-11-04-03-51-09-065\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------------!"
     ]
    }
   ],
   "source": [
    "rcf_predictor = rcf.deploy(\n",
    "    #endpoint_name='random-cut-forest-endpoint',\n",
    "    initial_instance_count=1,\n",
    "    instance_type='ml.p2.xlarge')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Deploy the model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Evaluate the model with test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "rcf_predictor.content_type = 'text/csv'\n",
    "rcf_predictor.serializer = csv_serializer\n",
    "rcf_predictor.accept = 'application/json'\n",
    "rcf_predictor.deserializer = json_deserializer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "def predict_rcf(current_predictor, data, rows=500):\n",
    "    split_array = np.array_split(data, int(data.shape[0] / float(rows) + 1))\n",
    "    predictions = []\n",
    "    for array in split_array:\n",
    "        array_preds = [s['score'] for s in current_predictor.predict(array)['scores']]\n",
    "        predictions.append(array_preds)\n",
    "\n",
    "    return np.concatenate([np.array(batch) for batch in predictions])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "positives = X_test[y_test == 1]\n",
    "positives_scores = predict_rcf(rcf_predictor, positives)\n",
    "\n",
    "negatives = X_test[y_test == 0]\n",
    "negatives_scores = predict_rcf(rcf_predictor, negatives)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2f50dac7ac8>"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "sns.distplot(positives_scores, label='fraud', bins=20)\n",
    "sns.distplot(negatives_scores, label='not-fraud', bins=20)\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The unsupervised model already can achieve some separation between the classes, with higher anomaly scores being correlated to fraud."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### (2) Supervised Learning"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Once we have gathered an ample amount of labeled training data, we can use a supervised learning algorithm that discovers relationships between the features and the dependent class.\n",
    "\n",
    "We will use **Gradient Boosted Trees** as our model, as they have a proven track record, are highly scalable and can deal with missing data, reducing the need to pre-process datasets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Upload data to S3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sklearn\n",
    "from sklearn.datasets import dump_svmlight_file   \n",
    "\n",
    "key = 'fraud-dataset'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Uploaded training data location: s3://sagemaker-us-west-2-570447867175/fraud-detec/fraud-dataset\n",
      "Training artifacts will be uploaded to: s3://sagemaker-us-west-2-570447867175/fraud-detec/output\n"
     ]
    }
   ],
   "source": [
    "# Convert data to Protobuf (binary stream) and upload to S3.\n",
    "def writeDatasetToProtobuf(bucket, prefix, key):\n",
    "    buf = io.BytesIO()\n",
    "    sklearn.datasets.dump_svmlight_file(X_train, y_train, buf)\n",
    "    buf.seek(0)\n",
    "    obj = '{}/{}'.format(prefix, key)\n",
    "    boto3.resource('s3').Bucket(bucket).Object(obj).upload_fileobj(buf)\n",
    "    return 's3://{}/{}'.format(bucket,obj)\n",
    "    \n",
    "    \n",
    "s3_train_data = writeDatasetToProtobuf(bucket, prefix, key)    \n",
    "print('Uploaded training data location: {}'.format(s3_train_data))\n",
    "\n",
    "output_location  = 's3://{}/{}/output'.format(bucket, prefix)\n",
    "print('Training artifacts will be uploaded to: {}'.format(output_location))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now train using SageMaker's built-in XGBoost algorithm. To specify the XGBoost algorithm, we use a utility function to obtain its URI. A complete list of built-in algorithms is found here: https://docs.aws.amazon.com/sagemaker/latest/dg/algos.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:sagemaker.amazon.amazon_estimator:'get_image_uri' method will be deprecated in favor of 'ImageURIProvider' class in SageMaker Python SDK v2.\n",
      "WARNING:root:There is a more up to date SageMaker XGBoost image. To use the newer image, please set 'repo_version'='1.0-1'. For example:\n",
      "\tget_image_uri(region, 'xgboost', '1.0-1').\n"
     ]
    }
   ],
   "source": [
    "from sagemaker.amazon.amazon_estimator import get_image_uri\n",
    "\n",
    "container = get_image_uri(region, \n",
    "                          'xgboost', \n",
    "                          #repo_version='0.90-2'\n",
    "                         )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Configure the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:Parameter image_name will be renamed to image_uri in SageMaker Python SDK v2.\n"
     ]
    }
   ],
   "source": [
    "from math import sqrt\n",
    "from sagemaker import get_execution_role\n",
    "\n",
    "# Because the data set is so highly skewed, we set the scale position weight conservatively,\n",
    "# as sqrt(num_nonfraud/num_fraud).\n",
    "# Other recommendations for the scale_pos_weight are setting it to (num_nonfraud/num_fraud).\n",
    "scale_pos_weight = sqrt(np.count_nonzero(y_train==0)/np.count_nonzero(y_train))\n",
    "\n",
    "hyperparameters = {\n",
    "    \"max_depth\":5,\n",
    "    \"subsample\":0.8,\n",
    "    \"num_round\":100,\n",
    "    \"eta\":0.2,\n",
    "    \"gamma\":4,\n",
    "    \"min_child_weight\":6,\n",
    "    \"silent\":0,\n",
    "    \"objective\":'binary:logistic',\n",
    "    \"eval_metric\":'auc',\n",
    "    \"scale_pos_weight\": scale_pos_weight\n",
    "}\n",
    "\n",
    "clf = sagemaker.estimator.Estimator(container,\n",
    "                                    role,\n",
    "                                    hyperparameters=hyperparameters,\n",
    "                                    train_instance_count=1, \n",
    "                                    train_instance_type='ml.m4.xlarge',\n",
    "                                    output_path=output_location,\n",
    "                                    sagemaker_session=sess)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:sagemaker:'s3_input' class will be renamed to 'TrainingInput' in SageMaker Python SDK v2.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-11-04 04:36:24 Starting - Starting the training job...\n",
      "2020-11-04 04:36:26 Starting - Launching requested ML instances......\n",
      "2020-11-04 04:37:55 Starting - Preparing the instances for training......\n",
      "2020-11-04 04:38:54 Downloading - Downloading input data...\n",
      "2020-11-04 04:39:27 Training - Downloading the training image..\n",
      "2020-11-04 04:39:46 Training - Training image download completed. Training in progress.\u001b[34mArguments: train\u001b[0m\n",
      "\u001b[34m[2020-11-04:04:39:47:INFO] Running standalone xgboost training.\u001b[0m\n",
      "\u001b[34m[2020-11-04:04:39:47:INFO] Path /opt/ml/input/data/validation does not exist!\u001b[0m\n",
      "\u001b[34m[2020-11-04:04:39:47:INFO] File size need to be processed in the node: 157.49mb. Available memory size in the node: 8470.21mb\u001b[0m\n",
      "\u001b[34m[04:39:47] S3DistributionType set as FullyReplicated\u001b[0m\n",
      "\u001b[34m[04:39:48] 256326x30 matrix with 7688169 entries loaded from /opt/ml/input/data/train\u001b[0m\n",
      "\u001b[34m[04:39:49] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 28 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[0]#011train-auc:0.95228\u001b[0m\n",
      "\u001b[34m[04:39:49] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 30 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[1]#011train-auc:0.961875\u001b[0m\n",
      "\u001b[34m[04:39:50] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[2]#011train-auc:0.961863\u001b[0m\n",
      "\u001b[34m[04:39:50] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[3]#011train-auc:0.963978\u001b[0m\n",
      "\u001b[34m[04:39:51] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 36 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[4]#011train-auc:0.968065\u001b[0m\n",
      "\u001b[34m[04:39:52] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 34 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[5]#011train-auc:0.973918\u001b[0m\n",
      "\u001b[34m[04:39:52] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 38 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[6]#011train-auc:0.974925\u001b[0m\n",
      "\u001b[34m[04:39:53] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 36 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[7]#011train-auc:0.985684\u001b[0m\n",
      "\u001b[34m[04:39:53] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 46 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[8]#011train-auc:0.988368\u001b[0m\n",
      "\u001b[34m[04:39:54] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 40 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[9]#011train-auc:0.988604\u001b[0m\n",
      "\u001b[34m[04:39:54] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 40 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[10]#011train-auc:0.988523\u001b[0m\n",
      "\u001b[34m[04:39:55] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 38 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[11]#011train-auc:0.991693\u001b[0m\n",
      "\u001b[34m[04:39:55] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 46 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[12]#011train-auc:0.991858\u001b[0m\n",
      "\u001b[34m[04:39:56] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 44 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[13]#011train-auc:0.992301\u001b[0m\n",
      "\u001b[34m[04:39:56] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 40 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[14]#011train-auc:0.993952\u001b[0m\n",
      "\u001b[34m[04:39:57] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 50 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[15]#011train-auc:0.995123\u001b[0m\n",
      "\u001b[34m[04:39:57] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 54 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[16]#011train-auc:0.995203\u001b[0m\n",
      "\u001b[34m[04:39:58] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 54 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[17]#011train-auc:0.996703\u001b[0m\n",
      "\u001b[34m[04:39:58] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 50 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[18]#011train-auc:0.997339\u001b[0m\n",
      "\u001b[34m[04:39:59] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 54 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[19]#011train-auc:0.99822\u001b[0m\n",
      "\u001b[34m[04:40:00] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 44 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[20]#011train-auc:0.999057\u001b[0m\n",
      "\u001b[34m[04:40:00] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 36 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[21]#011train-auc:0.999158\u001b[0m\n",
      "\u001b[34m[04:40:01] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 40 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[22]#011train-auc:0.999582\u001b[0m\n",
      "\u001b[34m[04:40:01] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 36 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[23]#011train-auc:0.99956\u001b[0m\n",
      "\u001b[34m[04:40:02] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 56 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[24]#011train-auc:0.999563\u001b[0m\n",
      "\u001b[34m[04:40:02] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 50 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[25]#011train-auc:0.99982\u001b[0m\n",
      "\u001b[34m[04:40:03] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 42 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[26]#011train-auc:0.999864\u001b[0m\n",
      "\u001b[34m[04:40:03] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 52 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[27]#011train-auc:0.999908\u001b[0m\n",
      "\u001b[34m[04:40:04] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 38 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[28]#011train-auc:0.999919\u001b[0m\n",
      "\u001b[34m[04:40:04] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 36 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[29]#011train-auc:0.999918\u001b[0m\n",
      "\u001b[34m[04:40:05] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 38 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[30]#011train-auc:0.99994\u001b[0m\n",
      "\u001b[34m[04:40:06] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 44 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[31]#011train-auc:0.999953\u001b[0m\n",
      "\u001b[34m[04:40:06] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 38 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[32]#011train-auc:0.99996\u001b[0m\n",
      "\u001b[34m[04:40:07] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 24 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[33]#011train-auc:0.999961\u001b[0m\n",
      "\u001b[34m[04:40:07] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 34 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[34]#011train-auc:0.999964\u001b[0m\n",
      "\u001b[34m[04:40:08] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 24 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[35]#011train-auc:0.999967\u001b[0m\n",
      "\u001b[34m[04:40:08] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 30 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[36]#011train-auc:0.99997\u001b[0m\n",
      "\u001b[34m[04:40:09] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 48 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[37]#011train-auc:0.999972\u001b[0m\n",
      "\u001b[34m[04:40:09] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[38]#011train-auc:0.999975\u001b[0m\n",
      "\u001b[34m[04:40:10] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 34 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[39]#011train-auc:0.999977\u001b[0m\n",
      "\u001b[34m[04:40:10] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 16 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[40]#011train-auc:0.999978\u001b[0m\n",
      "\u001b[34m[04:40:11] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 40 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[41]#011train-auc:0.99998\u001b[0m\n",
      "\u001b[34m[04:40:11] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[42]#011train-auc:0.999982\u001b[0m\n",
      "\u001b[34m[04:40:12] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 26 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[43]#011train-auc:0.999982\u001b[0m\n",
      "\u001b[34m[04:40:12] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 26 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[44]#011train-auc:0.999983\u001b[0m\n",
      "\u001b[34m[04:40:13] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 22 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[45]#011train-auc:0.999984\u001b[0m\n",
      "\u001b[34m[04:40:13] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 28 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[46]#011train-auc:0.999984\u001b[0m\n",
      "\u001b[34m[04:40:14] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 18 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[47]#011train-auc:0.999985\u001b[0m\n",
      "\u001b[34m[04:40:14] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[48]#011train-auc:0.999986\u001b[0m\n",
      "\u001b[34m[04:40:15] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 28 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[49]#011train-auc:0.999987\u001b[0m\n",
      "\u001b[34m[04:40:16] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 28 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[50]#011train-auc:0.999987\u001b[0m\n",
      "\u001b[34m[04:40:16] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 28 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[51]#011train-auc:0.999987\u001b[0m\n",
      "\u001b[34m[04:40:17] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 22 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[52]#011train-auc:0.999988\u001b[0m\n",
      "\u001b[34m[04:40:17] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 22 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[53]#011train-auc:0.999988\u001b[0m\n",
      "\u001b[34m[04:40:18] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 28 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[54]#011train-auc:0.999989\u001b[0m\n",
      "\u001b[34m[04:40:18] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 14 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[55]#011train-auc:0.999989\u001b[0m\n",
      "\u001b[34m[04:40:19] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 34 extra nodes, 10 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[56]#011train-auc:0.999991\u001b[0m\n",
      "\u001b[34m[04:40:19] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 26 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[57]#011train-auc:0.999991\u001b[0m\n",
      "\u001b[34m[04:40:20] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 12 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[58]#011train-auc:0.999991\u001b[0m\n",
      "\u001b[34m[04:40:20] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 18 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[59]#011train-auc:0.999991\u001b[0m\n",
      "\u001b[34m[04:40:21] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 22 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[60]#011train-auc:0.999992\u001b[0m\n",
      "\u001b[34m[04:40:21] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 28 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[61]#011train-auc:0.999992\u001b[0m\n",
      "\u001b[34m[04:40:22] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 24 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[62]#011train-auc:0.999993\u001b[0m\n",
      "\u001b[34m[04:40:22] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 16 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[63]#011train-auc:0.999993\u001b[0m\n",
      "\u001b[34m[04:40:23] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 22 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[64]#011train-auc:0.999993\u001b[0m\n",
      "\u001b[34m[04:40:23] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 14 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[65]#011train-auc:0.999994\u001b[0m\n",
      "\u001b[34m[04:40:24] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 22 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[66]#011train-auc:0.999994\u001b[0m\n",
      "\u001b[34m[04:40:24] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 16 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[67]#011train-auc:0.999995\u001b[0m\n",
      "\u001b[34m[04:40:25] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 20 extra nodes, 10 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[68]#011train-auc:0.999995\u001b[0m\n",
      "\u001b[34m[04:40:25] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 28 extra nodes, 14 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[69]#011train-auc:0.999995\u001b[0m\n",
      "\u001b[34m[04:40:26] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 26 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[70]#011train-auc:0.999996\u001b[0m\n",
      "\u001b[34m[04:40:26] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 12 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[71]#011train-auc:0.999996\u001b[0m\n",
      "\u001b[34m[04:40:27] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 26 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[72]#011train-auc:0.999996\u001b[0m\n",
      "\u001b[34m[04:40:27] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 10 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[73]#011train-auc:0.999996\u001b[0m\n",
      "\u001b[34m[04:40:28] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 18 extra nodes, 10 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[74]#011train-auc:0.999996\u001b[0m\n",
      "\u001b[34m[04:40:29] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 20 extra nodes, 10 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[75]#011train-auc:0.999997\u001b[0m\n",
      "\u001b[34m[04:40:29] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 14 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[76]#011train-auc:0.999996\u001b[0m\n",
      "\u001b[34m[04:40:30] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 16 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[77]#011train-auc:0.999997\u001b[0m\n",
      "\u001b[34m[04:40:30] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 16 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[78]#011train-auc:0.999997\u001b[0m\n",
      "\u001b[34m[04:40:31] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 14 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[79]#011train-auc:0.999997\u001b[0m\n",
      "\u001b[34m[04:40:31] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 14 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[80]#011train-auc:0.999997\u001b[0m\n",
      "\u001b[34m[04:40:32] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 18 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[81]#011train-auc:0.999998\u001b[0m\n",
      "\u001b[34m[04:40:32] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 10 extra nodes, 10 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[82]#011train-auc:0.999998\u001b[0m\n",
      "\u001b[34m[04:40:33] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 14 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[83]#011train-auc:0.999998\u001b[0m\n",
      "\u001b[34m[04:40:33] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 14 extra nodes, 12 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[84]#011train-auc:0.999998\u001b[0m\n",
      "\u001b[34m[04:40:34] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 16 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[85]#011train-auc:0.999998\u001b[0m\n",
      "\u001b[34m[04:40:34] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 12 extra nodes, 12 pruned nodes, max_depth=4\u001b[0m\n",
      "\u001b[34m[86]#011train-auc:0.999998\u001b[0m\n",
      "\u001b[34m[04:40:35] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 14 extra nodes, 12 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[87]#011train-auc:0.999998\u001b[0m\n",
      "\u001b[34m[04:40:35] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 2 extra nodes, 14 pruned nodes, max_depth=1\u001b[0m\n",
      "\u001b[34m[88]#011train-auc:0.999998\u001b[0m\n",
      "\u001b[34m[04:40:36] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 10 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[89]#011train-auc:0.999998\u001b[0m\n",
      "\u001b[34m[04:40:36] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 10 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[90]#011train-auc:0.999998\u001b[0m\n",
      "\u001b[34m[04:40:37] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 4 extra nodes, 8 pruned nodes, max_depth=2\u001b[0m\n",
      "\u001b[34m[91]#011train-auc:0.999998\u001b[0m\n",
      "\u001b[34m[04:40:37] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 16 extra nodes, 14 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[92]#011train-auc:0.999999\u001b[0m\n",
      "\u001b[34m[04:40:38] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 10 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[93]#011train-auc:0.999999\u001b[0m\n",
      "\u001b[34m[04:40:38] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 0 extra nodes, 14 pruned nodes, max_depth=0\u001b[0m\n",
      "\u001b[34m[94]#011train-auc:0.999999\u001b[0m\n",
      "\u001b[34m[04:40:38] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 12 extra nodes, 10 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[95]#011train-auc:0.999999\u001b[0m\n",
      "\u001b[34m[04:40:39] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 8 extra nodes, 10 pruned nodes, max_depth=4\u001b[0m\n",
      "\u001b[34m[96]#011train-auc:0.999999\u001b[0m\n",
      "\u001b[34m[04:40:39] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 14 extra nodes, 12 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[97]#011train-auc:0.999999\u001b[0m\n",
      "\u001b[34m[04:40:40] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 8 extra nodes, 8 pruned nodes, max_depth=4\u001b[0m\n",
      "\u001b[34m[98]#011train-auc:0.999999\u001b[0m\n",
      "\u001b[34m[04:40:40] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 14 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[99]#011train-auc:0.999999\u001b[0m\n",
      "\n",
      "2020-11-04 04:40:50 Uploading - Uploading generated training model\n",
      "2020-11-04 04:40:50 Completed - Training job completed\n",
      "Training seconds: 116\n",
      "Billable seconds: 116\n"
     ]
    }
   ],
   "source": [
    "clf.fit({'train': s3_train_data})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Deploy the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:sagemaker:Parameter image will be renamed to image_uri in SageMaker Python SDK v2.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------!"
     ]
    }
   ],
   "source": [
    "predictor = clf.deploy(initial_instance_count=1,\n",
    "                       endpoint_name=\"fraud-detection\",\n",
    "                       instance_type='ml.m4.xlarge', \n",
    "                       serializer=csv_serializer,\n",
    "                       deserializer=None,\n",
    "                       content_type='text/csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Evaluate the model with test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Because we have a large test set, we call predict on smaller batches\n",
    "def predict(current_predictor, data, rows=500):\n",
    "    split_array = np.array_split(data, int(data.shape[0] / float(rows) + 1))\n",
    "    predictions = ''\n",
    "    for array in split_array:\n",
    "        predictions = ','.join([predictions, current_predictor.predict(array).decode('utf-8')])\n",
    "\n",
    "    return np.fromstring(predictions[1:], sep=',')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "raw_preds = predict(predictor, X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.59439478e-05, 4.12945410e-05, 2.02579122e-05, ...,\n",
       "       1.71951797e-05, 1.62596807e-05, 5.44415816e-05])"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "raw_preds"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will use a few measures from the scikit-learn package to evaluate the performance of our model. When dealing with an imbalanced dataset, we need to choose metrics that take into account the frequency of each class in the data.\n",
    "\n",
    "Two such metrics are the [balanced accuracy score](https://scikit-learn.org/stable/modules/model_evaluation.html#balanced-accuracy-score), and [Cohen's Kappa](https://scikit-learn.org/stable/modules/model_evaluation.html#cohen-s-kappa)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Balanced accuracy = 0.9037758031479232\n",
      "Cohen's Kappa = 0.8568975801351875\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import balanced_accuracy_score, cohen_kappa_score\n",
    "\n",
    "# scikit-learn expects 0/1 predictions, so we threshold our raw predictions\n",
    "y_preds = np.where(raw_preds > 0.5, 1, 0)\n",
    "print(\"Balanced accuracy = {}\".format(balanced_accuracy_score(y_test, y_preds)))\n",
    "print(\"Cohen's Kappa = {}\".format(cohen_kappa_score(y_test, y_preds)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can already see that our model performs very well in terms of both metrics, Cohen's Kappa scores above 0.8 are generally very favorable.\n",
    "\n",
    "Apart from single-value metrics, it's also useful to look at metrics that indicate performance per class. A confusion matrix, and per-class precision, recall and f1-score can also provide more information about the model's performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "def plot_confusion_matrix(y_true, y_predicted):\n",
    "\n",
    "    cm  = confusion_matrix(y_true, y_predicted)\n",
    "    # Get the per-class normalized value for each cell\n",
    "    cm_norm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "    \n",
    "    # We color each cell according to its normalized value, annotate with exact counts.\n",
    "    ax = sns.heatmap(cm_norm, annot=cm, fmt=\"d\")\n",
    "    ax.set(xticklabels=[\"non-fraud\", \"fraud\"], yticklabels=[\"non-fraud\", \"fraud\"])\n",
    "    ax.set_ylim([0,2])\n",
    "    plt.title('Confusion Matrix')\n",
    "    plt.ylabel('Real Classes')\n",
    "    plt.xlabel('Predicted Classes')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "plot_confusion_matrix(y_test, y_preds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "   non-fraud       1.00      1.00      1.00     28429\n",
      "       fraud       0.91      0.81      0.86        52\n",
      "\n",
      "    accuracy                           1.00     28481\n",
      "   macro avg       0.96      0.90      0.93     28481\n",
      "weighted avg       1.00      1.00      1.00     28481\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import classification_report\n",
    "\n",
    "print(classification_report(\n",
    "    y_test, y_preds, target_names=['non-fraud', 'fraud']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4. Keep sending test traffic to the endpoint via lambda"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can now show how we could use both of these models in a production system, using HTTP requests to an AWS Lambda function that invokes both the unsupervised and the supervised SageMaker endpoints.\n",
    "\n",
    "We create a background thread that will constantly create HTTP requests to invoke the Lambda, using our test data as input. See the included generate_endpoint_traffic.py file to see how that is done. The output will be logged to Kinesis, and you can also observe it in the Lambda function's CloudWatch logs."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from threading import Thread\n",
    "from generate_endpoint_traffic import generate_traffic\n",
    "\n",
    "thread = Thread(target = generate_traffic, args=[np.copy(X_test)])\n",
    "thread.start()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5. Improve Class Imbalance in the dataset and re-train"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that we have a baseline model using XGBoost, we can try to see if sampling techniques that are designed specifically for imbalanced problems can improve the performance of the model.\n",
    "\n",
    "For that purpose we will be using the imbalanced-learn package that works well with scikit-learn. We have pre-installed the package for this kernel, but if you need it for a different Jupyter kernel you can install it by running pip install --upgrade imbalanced-learn within the conda environment you need.\n",
    "\n",
    "We will be using **Sythetic Minority Over-sampling (SMOTE)**, which oversamples the minority class by interpolating new data points between existing ones."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For a detailed information, please refer to this [documentation](https://imbalanced-learn.readthedocs.io/en/stable/auto_examples/over-sampling/plot_comparison_over_sampling.html#sphx-glr-auto-examples-over-sampling-plot-comparison-over-sampling-py)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "from imblearn.over_sampling import SMOTE\n",
    "\n",
    "smote = SMOTE(random_state=42)\n",
    "X_smote, y_smote = smote.fit_resample(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[(0.0, 255886), (1.0, 255886)]\n"
     ]
    }
   ],
   "source": [
    "from collections import Counter\n",
    "print(sorted(Counter(y_smote).items()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We note that this is a case of extreme oversampling of the the minority class, we went from ~0.17% to 50%. An alternative would be to use a smaller resampling ratio, such as having one minority cl ass sample for every sqrt(non_fraud/fraud) majority samples, or using more advanced resampling techniques. See the comparison provided by imbalanced-learn for more over-sampling options.\n",
    "\n",
    "In our case we'll use the SMOTE dataset we just created and upload it to S3 for training."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Uploaded training data location: s3://sagemaker-us-west-2-570447867175/fraud-detec/fraud-dataset\n",
      "Training artifacts will be uploaded to: s3://sagemaker-us-west-2-570447867175/fraud-detec/output\n"
     ]
    }
   ],
   "source": [
    "# Convert data to Protobuf (binary stream) and upload to S3.\n",
    "def writeDatasetToProtobuf_smote(bucket, prefix, key):\n",
    "    smote_buf = io.BytesIO()\n",
    "    \n",
    "    # Dump the SMOTE data into a buffer\n",
    "    sklearn.datasets.dump_svmlight_file(X_smote, y_smote, smote_buf)\n",
    "    smote_buf.seek(0)\n",
    "    obj = '{}/{}'.format(prefix, key)\n",
    "    \n",
    "    # Upload from the buffer to S3\n",
    "    boto3.resource('s3').Bucket(bucket).Object(obj).upload_fileobj(smote_buf)\n",
    "    return 's3://{}/{}'.format(bucket,obj)\n",
    "    \n",
    "    \n",
    "s3_smote_train_data = writeDatasetToProtobuf_smote(bucket, prefix, key)    \n",
    "print('Uploaded training data location: {}'.format(s3_smote_train_data))\n",
    "\n",
    "smote_output_location  = 's3://{}/{}/output'.format(bucket, prefix)\n",
    "print('Training artifacts will be uploaded to: {}'.format(smote_output_location))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:Parameter image_name will be renamed to image_uri in SageMaker Python SDK v2.\n"
     ]
    }
   ],
   "source": [
    "try:\n",
    "    hyperparameters.pop(\"scale_pos_weight\") # No need to scale weights after SMOTE resampling, so we remove that parameter\n",
    "except KeyError:\n",
    "    # In case the parameter was already removed, do nothing.\n",
    "    pass\n",
    "smote_xgb = sagemaker.estimator.Estimator(container,\n",
    "                                        role,\n",
    "                                        hyperparameters=hyperparameters,\n",
    "                                        train_instance_count=1, \n",
    "                                        train_instance_type='ml.m4.xlarge',\n",
    "                                        output_path=smote_output_location,\n",
    "                                        sagemaker_session=sess)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Training with smote-d dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:sagemaker:'s3_input' class will be renamed to 'TrainingInput' in SageMaker Python SDK v2.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2020-11-04 05:11:51 Starting - Starting the training job...\n",
      "2020-11-04 05:11:53 Starting - Launching requested ML instances......\n",
      "2020-11-04 05:13:21 Starting - Preparing the instances for training.........\n",
      "2020-11-04 05:14:32 Downloading - Downloading input data...\n",
      "2020-11-04 05:15:28 Training - Training image download completed. Training in progress..\u001b[34mArguments: train\u001b[0m\n",
      "\u001b[34m[2020-11-04:05:15:29:INFO] Running standalone xgboost training.\u001b[0m\n",
      "\u001b[34m[2020-11-04:05:15:29:INFO] Path /opt/ml/input/data/validation does not exist!\u001b[0m\n",
      "\u001b[34m[2020-11-04:05:15:29:INFO] File size need to be processed in the node: 314.05mb. Available memory size in the node: 8469.27mb\u001b[0m\n",
      "\u001b[34m[05:15:29] S3DistributionType set as FullyReplicated\u001b[0m\n",
      "\u001b[34m[05:15:30] 511772x30 matrix with 15350375 entries loaded from /opt/ml/input/data/train\u001b[0m\n",
      "\u001b[34m[05:15:31] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 52 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[0]#011train-auc:0.991899\u001b[0m\n",
      "\u001b[34m[05:15:32] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 54 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[1]#011train-auc:0.993737\u001b[0m\n",
      "\u001b[34m[05:15:34] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 54 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[2]#011train-auc:0.994626\u001b[0m\n",
      "\u001b[34m[05:15:35] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 58 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[3]#011train-auc:0.995512\u001b[0m\n",
      "\u001b[34m[05:15:36] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 56 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[4]#011train-auc:0.996697\u001b[0m\n",
      "\u001b[34m[05:15:37] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 58 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[5]#011train-auc:0.996985\u001b[0m\n",
      "\u001b[34m[05:15:38] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 62 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[6]#011train-auc:0.997448\u001b[0m\n",
      "\u001b[34m[05:15:40] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 56 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[7]#011train-auc:0.997586\u001b[0m\n",
      "\u001b[34m[05:15:41] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 56 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[8]#011train-auc:0.99778\u001b[0m\n",
      "\u001b[34m[05:15:42] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 56 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[9]#011train-auc:0.997943\u001b[0m\n",
      "\u001b[34m[05:15:43] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 56 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[10]#011train-auc:0.998065\u001b[0m\n",
      "\u001b[34m[05:15:44] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 56 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[11]#011train-auc:0.998498\u001b[0m\n",
      "\u001b[34m[05:15:45] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 52 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[12]#011train-auc:0.998671\u001b[0m\n",
      "\u001b[34m[05:15:46] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 54 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[13]#011train-auc:0.998793\u001b[0m\n",
      "\u001b[34m[05:15:47] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 50 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[14]#011train-auc:0.998955\u001b[0m\n",
      "\u001b[34m[05:15:48] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 48 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[15]#011train-auc:0.999052\u001b[0m\n",
      "\u001b[34m[05:15:49] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 50 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[16]#011train-auc:0.999175\u001b[0m\n",
      "\u001b[34m[05:15:50] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 48 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[17]#011train-auc:0.999275\u001b[0m\n",
      "\u001b[34m[05:15:51] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 50 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[18]#011train-auc:0.999344\u001b[0m\n",
      "\u001b[34m[05:15:52] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 42 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[19]#011train-auc:0.999396\u001b[0m\n",
      "\u001b[34m[05:15:54] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 48 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[20]#011train-auc:0.999425\u001b[0m\n",
      "\u001b[34m[05:15:55] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 48 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[21]#011train-auc:0.999498\u001b[0m\n",
      "\u001b[34m[05:15:56] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 42 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[22]#011train-auc:0.999573\u001b[0m\n",
      "\u001b[34m[05:15:57] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 52 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[23]#011train-auc:0.999631\u001b[0m\n",
      "\u001b[34m[05:15:58] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 44 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[24]#011train-auc:0.999665\u001b[0m\n",
      "\u001b[34m[05:15:59] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 40 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[25]#011train-auc:0.999689\u001b[0m\n",
      "\u001b[34m[05:16:00] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 40 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[26]#011train-auc:0.999706\u001b[0m\n",
      "\u001b[34m[05:16:01] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 38 extra nodes, 12 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[27]#011train-auc:0.99973\u001b[0m\n",
      "\u001b[34m[05:16:02] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 50 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[28]#011train-auc:0.999751\u001b[0m\n",
      "\u001b[34m[05:16:03] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 52 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[29]#011train-auc:0.999801\u001b[0m\n",
      "\u001b[34m[05:16:04] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 56 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[30]#011train-auc:0.999825\u001b[0m\n",
      "\u001b[34m[05:16:05] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[31]#011train-auc:0.999833\u001b[0m\n",
      "\u001b[34m[05:16:07] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 36 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[32]#011train-auc:0.999842\u001b[0m\n",
      "\u001b[34m[05:16:08] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 52 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[33]#011train-auc:0.999852\u001b[0m\n",
      "\u001b[34m[05:16:09] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 48 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[34]#011train-auc:0.999863\u001b[0m\n",
      "\u001b[34m[05:16:10] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 40 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[35]#011train-auc:0.999868\u001b[0m\n",
      "\u001b[34m[05:16:11] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 30 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[36]#011train-auc:0.999895\u001b[0m\n",
      "\u001b[34m[05:16:12] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 28 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[37]#011train-auc:0.999898\u001b[0m\n",
      "\u001b[34m[05:16:13] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 44 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[38]#011train-auc:0.999908\u001b[0m\n",
      "\u001b[34m[05:16:14] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 14 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[39]#011train-auc:0.999909\u001b[0m\n",
      "\u001b[34m[05:16:15] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 56 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[40]#011train-auc:0.999918\u001b[0m\n",
      "\u001b[34m[05:16:16] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 46 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[41]#011train-auc:0.999921\u001b[0m\n",
      "\u001b[34m[05:16:17] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 44 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[42]#011train-auc:0.999926\u001b[0m\n",
      "\u001b[34m[05:16:18] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 42 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[43]#011train-auc:0.999931\u001b[0m\n",
      "\u001b[34m[05:16:19] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 18 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[44]#011train-auc:0.999933\u001b[0m\n",
      "\u001b[34m[05:16:20] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 42 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[45]#011train-auc:0.999937\u001b[0m\n",
      "\u001b[34m[05:16:22] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[46]#011train-auc:0.999939\u001b[0m\n",
      "\u001b[34m[05:16:23] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 46 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[47]#011train-auc:0.999941\u001b[0m\n",
      "\u001b[34m[05:16:24] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 54 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[48]#011train-auc:0.999947\u001b[0m\n",
      "\u001b[34m[05:16:25] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 44 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[49]#011train-auc:0.999948\u001b[0m\n",
      "\u001b[34m[05:16:26] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 50 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[50]#011train-auc:0.999954\u001b[0m\n",
      "\u001b[34m[05:16:27] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 40 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[51]#011train-auc:0.999955\u001b[0m\n",
      "\u001b[34m[05:16:28] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 30 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[52]#011train-auc:0.999961\u001b[0m\n",
      "\u001b[34m[05:16:29] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 46 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[53]#011train-auc:0.999963\u001b[0m\n",
      "\u001b[34m[05:16:30] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 24 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[54]#011train-auc:0.999964\u001b[0m\n",
      "\u001b[34m[05:16:31] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 26 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[55]#011train-auc:0.999964\u001b[0m\n",
      "\u001b[34m[05:16:32] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 40 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[56]#011train-auc:0.999967\u001b[0m\n",
      "\u001b[34m[05:16:33] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 46 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[57]#011train-auc:0.99997\u001b[0m\n",
      "\u001b[34m[05:16:34] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[58]#011train-auc:0.99997\u001b[0m\n",
      "\u001b[34m[05:16:35] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 50 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[59]#011train-auc:0.999972\u001b[0m\n",
      "\u001b[34m[05:16:36] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 46 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[60]#011train-auc:0.999973\u001b[0m\n",
      "\u001b[34m[05:16:38] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 40 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[61]#011train-auc:0.999974\u001b[0m\n",
      "\u001b[34m[05:16:39] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 34 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[62]#011train-auc:0.999976\u001b[0m\n",
      "\u001b[34m[05:16:40] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 30 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[63]#011train-auc:0.999977\u001b[0m\n",
      "\u001b[34m[05:16:41] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 50 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[64]#011train-auc:0.999977\u001b[0m\n",
      "\u001b[34m[05:16:42] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 24 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[65]#011train-auc:0.999979\u001b[0m\n",
      "\u001b[34m[05:16:43] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 38 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[66]#011train-auc:0.999981\u001b[0m\n",
      "\u001b[34m[05:16:44] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 20 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[67]#011train-auc:0.999981\u001b[0m\n",
      "\u001b[34m[05:16:45] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 34 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[68]#011train-auc:0.999981\u001b[0m\n",
      "\u001b[34m[05:16:46] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[69]#011train-auc:0.999981\u001b[0m\n",
      "\u001b[34m[05:16:47] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 38 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[70]#011train-auc:0.999983\u001b[0m\n",
      "\u001b[34m[05:16:48] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 34 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[71]#011train-auc:0.999984\u001b[0m\n",
      "\u001b[34m[05:16:49] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 36 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[72]#011train-auc:0.999985\u001b[0m\n",
      "\u001b[34m[05:16:50] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 28 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[73]#011train-auc:0.999986\u001b[0m\n",
      "\u001b[34m[05:16:51] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 38 extra nodes, 10 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[74]#011train-auc:0.999987\u001b[0m\n",
      "\u001b[34m[05:16:52] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[75]#011train-auc:0.999987\u001b[0m\n",
      "\u001b[34m[05:16:53] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 24 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[76]#011train-auc:0.999987\u001b[0m\n",
      "\u001b[34m[05:16:54] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 46 extra nodes, 10 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[77]#011train-auc:0.999988\u001b[0m\n",
      "\u001b[34m[05:16:55] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 36 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[78]#011train-auc:0.999989\u001b[0m\n",
      "\u001b[34m[05:16:56] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 34 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[79]#011train-auc:0.99999\u001b[0m\n",
      "\u001b[34m[05:16:57] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 38 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[80]#011train-auc:0.99999\u001b[0m\n",
      "\u001b[34m[05:16:58] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 30 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[81]#011train-auc:0.99999\u001b[0m\n",
      "\u001b[34m[05:16:59] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 20 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[82]#011train-auc:0.999991\u001b[0m\n",
      "\u001b[34m[05:17:01] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 30 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[83]#011train-auc:0.999991\u001b[0m\n",
      "\u001b[34m[05:17:02] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[84]#011train-auc:0.999991\u001b[0m\n",
      "\u001b[34m[05:17:03] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 36 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[85]#011train-auc:0.999992\u001b[0m\n",
      "\u001b[34m[05:17:04] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 34 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[86]#011train-auc:0.999992\u001b[0m\n",
      "\u001b[34m[05:17:05] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 40 extra nodes, 2 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[87]#011train-auc:0.999992\u001b[0m\n",
      "\u001b[34m[05:17:06] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[88]#011train-auc:0.999993\u001b[0m\n",
      "\u001b[34m[05:17:07] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 30 extra nodes, 8 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[89]#011train-auc:0.999993\u001b[0m\n",
      "\u001b[34m[05:17:08] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 26 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[90]#011train-auc:0.999994\u001b[0m\n",
      "\u001b[34m[05:17:09] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 42 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[91]#011train-auc:0.999994\u001b[0m\n",
      "\u001b[34m[05:17:10] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[92]#011train-auc:0.999994\u001b[0m\n",
      "\u001b[34m[05:17:11] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 30 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[93]#011train-auc:0.999995\u001b[0m\n",
      "\u001b[34m[05:17:12] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 32 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[94]#011train-auc:0.999995\u001b[0m\n",
      "\u001b[34m[05:17:13] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 26 extra nodes, 0 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[95]#011train-auc:0.999995\u001b[0m\n",
      "\u001b[34m[05:17:14] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 20 extra nodes, 4 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[96]#011train-auc:0.999996\u001b[0m\n",
      "\u001b[34m[05:17:15] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 42 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[97]#011train-auc:0.999996\u001b[0m\n",
      "\u001b[34m[05:17:16] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 24 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[98]#011train-auc:0.999996\u001b[0m\n",
      "\u001b[34m[05:17:17] src/tree/updater_prune.cc:74: tree pruning end, 1 roots, 24 extra nodes, 6 pruned nodes, max_depth=5\u001b[0m\n",
      "\u001b[34m[99]#011train-auc:0.999996\u001b[0m\n",
      "\n",
      "2020-11-04 05:17:27 Uploading - Uploading generated training model\n",
      "2020-11-04 05:17:27 Completed - Training job completed\n",
      "Training seconds: 175\n",
      "Billable seconds: 175\n"
     ]
    }
   ],
   "source": [
    "smote_xgb.fit({'train': s3_smote_train_data})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Deploying the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:sagemaker:Parameter image will be renamed to image_uri in SageMaker Python SDK v2.\n",
      "WARNING:sagemaker:Using already existing model: xgboost-2020-11-04-05-11-51-326\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-------------!"
     ]
    }
   ],
   "source": [
    "smote_predictor = smote_xgb.deploy(initial_instance_count=1,\n",
    "                       endpoint_name=\"fraud-detection-smote-endpoint\",\n",
    "                       instance_type='ml.m4.xlarge')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Specify input and output formats.\n",
    "smote_predictor.content_type = 'text/csv'\n",
    "smote_predictor.serializer = csv_serializer\n",
    "smote_predictor.deserializer = None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "smote_raw_preds = predict(smote_predictor, X_test)\n",
    "smote_preds = np.where(smote_raw_preds > 0.5, 1, 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Balanced accuracy = 0.9130394342721544\n",
      "Cohen's Kappa = 0.7221177723488154\n"
     ]
    }
   ],
   "source": [
    "print(\"Balanced accuracy = {}\".format(balanced_accuracy_score(y_test, smote_preds)))\n",
    "print(\"Cohen's Kappa = {}\".format(cohen_kappa_score(y_test, smote_preds)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "plot_confusion_matrix(y_test, smote_preds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "   non-fraud       1.00      1.00      1.00     28429\n",
      "       fraud       0.64      0.83      0.72        52\n",
      "\n",
      "    accuracy                           1.00     28481\n",
      "   macro avg       0.82      0.91      0.86     28481\n",
      "weighted avg       1.00      1.00      1.00     28481\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(\n",
    "    y_test, smote_preds, target_names=['non-fraud', 'fraud']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Due to the randomness of XGBoost your results may vary, but overall, you should see a large increase in non-fraud cases being classified as fraud (false positives). The reason this happens is because SMOTE has oversampled the fraud class so much that it's increased its overlap in feature space with the non-fraud cases. Since Cohen's Kappa gives more weight to false positives than balanced accuracy does, the metric drops significantly, as does the precision and F1 score for fraud cases. However, we can bring a balance between the metrics again by adjusting our classification threshold.\n",
    "\n",
    "So far we've been using 0.5 as the threshold between labeling a point as fraud or not. We can try different thresholds to see if they affect the result of the classification. To evaluate we'll use the balanced accuracy and Cohen's Kappa metrics."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Threshold: 0.1\n",
      "Balanced accuracy = 0.921\n",
      "Cohen's Kappa = 0.434\n",
      "\n",
      "Threshold: 0.2\n",
      "Balanced accuracy = 0.922\n",
      "Cohen's Kappa = 0.598\n",
      "\n",
      "Threshold: 0.3\n",
      "Balanced accuracy = 0.922\n",
      "Cohen's Kappa = 0.661\n",
      "\n",
      "Threshold: 0.4\n",
      "Balanced accuracy = 0.913\n",
      "Cohen's Kappa = 0.699\n",
      "\n",
      "Threshold: 0.5\n",
      "Balanced accuracy = 0.913\n",
      "Cohen's Kappa = 0.722\n",
      "\n",
      "Threshold: 0.6\n",
      "Balanced accuracy = 0.913\n",
      "Cohen's Kappa = 0.747\n",
      "\n",
      "Threshold: 0.7\n",
      "Balanced accuracy = 0.913\n",
      "Cohen's Kappa = 0.796\n",
      "\n",
      "Threshold: 0.8\n",
      "Balanced accuracy = 0.913\n",
      "Cohen's Kappa = 0.827\n",
      "\n",
      "Threshold: 0.9\n",
      "Balanced accuracy = 0.913\n",
      "Cohen's Kappa = 0.851\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for thres in np.linspace(0.1, 0.9, num=9):\n",
    "    smote_thres_preds = np.where(smote_raw_preds > thres, 1, 0)\n",
    "    print(\"Threshold: {:.1f}\".format(thres))\n",
    "    print(\"Balanced accuracy = {:.3f}\".format(balanced_accuracy_score(y_test, smote_thres_preds)))\n",
    "    print(\"Cohen's Kappa = {:.3f}\\n\".format(cohen_kappa_score(y_test, smote_thres_preds)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We see that Cohen's Kappa keeps increasing along with the threshold, without a significant loss in balanced accuracy. This adds a useful knob to our model: We can keep a low threshold if we care more about not missing any fraudulent cases, or we can increase the threshold to try to minimize the number of false positives."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 13. Close the SageMaker Instance"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will leave the unsupervised and base XGBoost endpoints running at the end of this notebook so we can handle incoming event streams. However, don't forget to delete the prediction endpoints when you're done. You can do that at the Amazon SageMaker console in the Endpoints page. Or you can run predictor_name.delete_endpoint()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Uncomment below to clean up endpoints\n",
    "#rcf_predictor.delete_endpoint()\n",
    "#predictor.delete_endpoint()\n",
    "smote_predictor.delete_endpoint()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Reference\n",
    ">https://aws.amazon.com/solutions/implementations/fraud-detection-using-machine-learning/"
   ]
  }
 ],
 "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}