azureml/azureml-snowflake-snowpark-sample.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "# The notebook connects gets data from Snowflake to train a ML model and deploy it on Snowpark to run batch inference.\r\n",
        "\r\n",
        "@author Mash Syed, Principal CSA, Microsoft RCG, 2022 Snowflake Data Superhero\r\n",
        "\r\n",
        "@license Apache License, Version 2.0\r\n",
        "### Dataset ####\r\n",
        "#### https://www.kaggle.com/datasets/neelesh0602/bankcsv\r\n"
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#pip install snowflake-connector-python\r\n",
        "#pip install \"snowflake-connector-python[pandas]\"\r\n",
        "#pip install --upgrade snowflake-sqlalchemy\r\n",
        "#pip install xgboost==1.5.0\r\n",
        "#pip install snowflake-snowpark-python\r\n",
        "#pip install \"snowflake-snowpark-python[pandas]\""
      ],
      "outputs": [],
      "execution_count": 143,
      "metadata": {
        "application/vnd.databricks.v1+cell": {
          "title": "",
          "showTitle": false,
          "inputWidgets": {},
          "nuid": "0d6e8d28-0856-4161-a5f5-64729de232ce"
        },
        "gather": {
          "logged": 1658347727340
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# azure ml packages\n",
        "from azureml.core import Workspace\n",
        "from azureml.core import Keyvault\n",
        "\n",
        "# snowflake packages\n",
        "from snowflake.snowpark import Session\n",
        "from snowflake.snowpark.functions import udf\n",
        "from snowflake.snowpark.types import *\n",
        "from snowflake.snowpark import functions as F\n",
        "\n",
        "# import standard python libs\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import snowflake.connector\n",
        "%matplotlib inline\n",
        "\n",
        "# Pull in relevant ML libraries\n",
        "from sklearn.metrics import confusion_matrix\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.model_selection import cross_val_score, KFold\n",
        "from sklearn.preprocessing import OneHotEncoder\n",
        "from sklearn.preprocessing import MinMaxScaler\n",
        "\n",
        "#Classifier\n",
        "from sklearn.ensemble import RandomForestClassifier\n",
        "\n",
        "# use sqlalchemy connector to write to snowflake\n",
        "from sqlalchemy.dialects import registry\n",
        "registry.register('snowflake', 'snowflake.sqlalchemy', 'dialect')\n",
        "from snowflake.sqlalchemy import URL\n",
        "from sqlalchemy import create_engine"
      ],
      "outputs": [],
      "execution_count": 144,
      "metadata": {
        "application/vnd.databricks.v1+cell": {
          "title": "",
          "showTitle": false,
          "inputWidgets": {},
          "nuid": "b266dda4-b4ed-4cdc-999c-2dc253860656"
        },
        "gather": {
          "logged": 1658291885363
        },
        "ms_comments": [
          {
            "threadId": "12d8fca3-f801-49b1-b6ae-c6bf82550282",
            "text": "Snowpark supports XGBBoost too. Check this link https://medium.com/snowflake/using-snowpark-for-python-with-amazon-sagemaker-44ec7fdb4381\n\n\nWe can remove this package during cleanup",
            "status": "resolved",
            "user": {
              "name": "Shankar Narayanan Ganapathy",
              "idType": "aad"
            },
            "createdDateUTC": 1660067231095,
            "modifiedDateUTC": 1660328323050,
            "replies": []
          },
          {
            "threadId": "f397b8f9-b3d9-47de-9c8c-9b50e481d6a7",
            "text": "Do we need Sqlalchemy for this? Cant we use just Snowflake Python connector?",
            "status": "resolved",
            "user": {
              "name": "Shankar Narayanan Ganapathy",
              "idType": "aad"
            },
            "createdDateUTC": 1660067145589,
            "modifiedDateUTC": 1660328327803,
            "replies": [
              {
                "replyId": "012e90ff-efac-4907-8cc9-0509b614c6ff",
                "text": "or the Snowflake Pandas connector",
                "user": {
                  "name": "Shankar Narayanan Ganapathy",
                  "idType": "aad"
                },
                "createdDateUTC": 1660067803353,
                "modifiedDateUTC": 1660067803353
              }
            ]
          }
        ],
        "ms_comment_ranges": {
          "f397b8f9-b3d9-47de-9c8c-9b50e481d6a7": {
            "text": "snowflake",
            "start": {
              "line": 28,
              "column": 40
            },
            "end": {
              "line": 28,
              "column": 49
            }
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Reading credentials from secrets\r\n",
        "ws = Workspace.from_config()\r\n",
        "\r\n",
        "keyvault = ws.get_default_keyvault()\r\n",
        "sfuser = keyvault.get_secret(name=\"sf-user\")\r\n",
        "sfpasswd = keyvault.get_secret(name=\"sf-passwd\")\r\n",
        "sfaccount = keyvault.get_secret(name=\"sf-account\")\r\n",
        "sfhost = keyvault.get_secret(name=\"sf-host\")\r\n",
        "sfhostalchemy = keyvault.get_secret(name='sf-host-alchemy')"
      ],
      "outputs": [],
      "execution_count": 145,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Connect To Snowflake and Get Data\r\n",
        "#### The Snowflake Python connector is used to connect to Snowflake to fetch the data into a pandas data frame.\r\n",
        "#### Reference - https://docs.snowflake.com/en/user-guide/python-connector.html\r\n"
      ],
      "metadata": {
        "application/vnd.databricks.v1+cell": {
          "title": "",
          "showTitle": false,
          "inputWidgets": {},
          "nuid": "b6b2a9c2-2ef0-4436-9e51-f78f55d9d5f4"
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "ctx = snowflake.connector.connect(\n",
        "    host=sfhost,\n",
        "    user=sfuser,\n",
        "    password=sfpasswd,\n",
        "    account=sfaccount,\n",
        "    warehouse=\"compute_wh\",\n",
        "    database=\"demo\",\n",
        "    schema=\"source_data\"\n",
        ")\n",
        "\n",
        "# Create a cursor object.\n",
        "cur = ctx.cursor()\n",
        "\n",
        "# Execute a statement that will generate a result set.\n",
        "sql = \"select * from demo.source_data.banking_customers\"\n",
        "cur.execute(sql)\n",
        "\n",
        "# Fetch the result set from the cursor and deliver it as the Pandas DataFrame.\n",
        "df = cur.fetch_pandas_all()\n",
        "\n",
        "df.head()"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 146,
          "data": {
            "text/plain": "   AGE         JOB  MARITAL  EDUCATION DEFAULT  BALANCE HOUSING LOAN  CONTACT  \\\n0   59      admin.  married  secondary      no     2343     yes   no  unknown   \n1   56      admin.  married  secondary      no       45      no   no  unknown   \n2   41  technician  married  secondary      no     1270     yes   no  unknown   \n3   55    services  married  secondary      no     2476     yes   no  unknown   \n4   54      admin.  married   tertiary      no      184      no   no  unknown   \n\n  DAY MONTH  DURATION CAMPAIGN  PDAYS  PREVIOUS POUTCOME DEPOSIT  CUSTOMERID  \n0   5   may      1042        1     -1         0  unknown     yes           1  \n1   5   may      1467        1     -1         0  unknown     yes           2  \n2   5   may      1389        1     -1         0  unknown     yes           3  \n3   5   may       579        1     -1         0  unknown     yes           4  \n4   5   may       673        2     -1         0  unknown     yes           5  ",
            "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>AGE</th>\n      <th>JOB</th>\n      <th>MARITAL</th>\n      <th>EDUCATION</th>\n      <th>DEFAULT</th>\n      <th>BALANCE</th>\n      <th>HOUSING</th>\n      <th>LOAN</th>\n      <th>CONTACT</th>\n      <th>DAY</th>\n      <th>MONTH</th>\n      <th>DURATION</th>\n      <th>CAMPAIGN</th>\n      <th>PDAYS</th>\n      <th>PREVIOUS</th>\n      <th>POUTCOME</th>\n      <th>DEPOSIT</th>\n      <th>CUSTOMERID</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>59</td>\n      <td>admin.</td>\n      <td>married</td>\n      <td>secondary</td>\n      <td>no</td>\n      <td>2343</td>\n      <td>yes</td>\n      <td>no</td>\n      <td>unknown</td>\n      <td>5</td>\n      <td>may</td>\n      <td>1042</td>\n      <td>1</td>\n      <td>-1</td>\n      <td>0</td>\n      <td>unknown</td>\n      <td>yes</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>56</td>\n      <td>admin.</td>\n      <td>married</td>\n      <td>secondary</td>\n      <td>no</td>\n      <td>45</td>\n      <td>no</td>\n      <td>no</td>\n      <td>unknown</td>\n      <td>5</td>\n      <td>may</td>\n      <td>1467</td>\n      <td>1</td>\n      <td>-1</td>\n      <td>0</td>\n      <td>unknown</td>\n      <td>yes</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>41</td>\n      <td>technician</td>\n      <td>married</td>\n      <td>secondary</td>\n      <td>no</td>\n      <td>1270</td>\n      <td>yes</td>\n      <td>no</td>\n      <td>unknown</td>\n      <td>5</td>\n      <td>may</td>\n      <td>1389</td>\n      <td>1</td>\n      <td>-1</td>\n      <td>0</td>\n      <td>unknown</td>\n      <td>yes</td>\n      <td>3</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>55</td>\n      <td>services</td>\n      <td>married</td>\n      <td>secondary</td>\n      <td>no</td>\n      <td>2476</td>\n      <td>yes</td>\n      <td>no</td>\n      <td>unknown</td>\n      <td>5</td>\n      <td>may</td>\n      <td>579</td>\n      <td>1</td>\n      <td>-1</td>\n      <td>0</td>\n      <td>unknown</td>\n      <td>yes</td>\n      <td>4</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>54</td>\n      <td>admin.</td>\n      <td>married</td>\n      <td>tertiary</td>\n      <td>no</td>\n      <td>184</td>\n      <td>no</td>\n      <td>no</td>\n      <td>unknown</td>\n      <td>5</td>\n      <td>may</td>\n      <td>673</td>\n      <td>2</td>\n      <td>-1</td>\n      <td>0</td>\n      <td>unknown</td>\n      <td>yes</td>\n      <td>5</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
          },
          "metadata": {}
        }
      ],
      "execution_count": 146,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        },
        "gather": {
          "logged": 1658348554540
        },
        "ms_comments": [
          {
            "threadId": "17e08237-bd5d-42a8-a089-c48c40f355e7",
            "text": "I will change this to Keypair after the notebook is completed",
            "status": "resolved",
            "user": {
              "name": "Shankar Narayanan Ganapathy",
              "idType": "aad"
            },
            "createdDateUTC": 1660067277142,
            "modifiedDateUTC": 1661759158405,
            "replies": []
          }
        ],
        "ms_comment_ranges": {
          "17e08237-bd5d-42a8-a089-c48c40f355e7": {
            "text": "connect",
            "start": {
              "line": 1,
              "column": 27
            },
            "end": {
              "line": 1,
              "column": 34
            }
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# The first thing we should do is convert our target from categorical to a binary value\r\n",
        "#### This allows us to group by the target and better understand how the features relate to it.\r\n"
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "## map the target variable to 1 or 0\r\n",
        "df['DEFAULT_STATUS'] = np.where(df.DEFAULT == 'yes', 1, 0)\r\n",
        "## drop the original column\r\n",
        "df.drop(columns='DEFAULT', inplace=True)"
      ],
      "outputs": [],
      "execution_count": 147,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Exploratory Data Analysis\r\n",
        "#### Exlploratory Data Analysis (EDA) is a process that data scientists apply to investigate, profile, test hypothesis, summarize, and visualize a dataset. In our dataset, we apply the following EDA methods:\r\n",
        "\r\n",
        "- Use summary statistics to describe the dataframe\r\n",
        "- Check for class imbalance\r\n",
        "- Summarize data based on multiple variables\r\n",
        "- Visualize using Bar Charts & Histograms\r\n",
        "- Correlation Matrix"
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Summary Statistics"
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df.describe()"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 148,
          "data": {
            "text/plain": "                AGE       BALANCE      DURATION         PDAYS      PREVIOUS  \\\ncount  11162.000000  11162.000000  11162.000000  11162.000000  11162.000000   \nmean      41.231948   1528.538524    371.993818     51.330407      0.832557   \nstd       11.913369   3225.413326    347.128386    108.758282      2.292007   \nmin       18.000000  -6847.000000      2.000000     -1.000000      0.000000   \n25%       32.000000    122.000000    138.000000     -1.000000      0.000000   \n50%       39.000000    550.000000    255.000000     -1.000000      0.000000   \n75%       49.000000   1708.000000    496.000000     20.750000      1.000000   \nmax       95.000000  81204.000000   3881.000000    854.000000     58.000000   \n\n         CUSTOMERID  DEFAULT_STATUS  \ncount  11162.000000    11162.000000  \nmean    5581.500000        0.015051  \nstd     3222.336187        0.121761  \nmin        1.000000        0.000000  \n25%     2791.250000        0.000000  \n50%     5581.500000        0.000000  \n75%     8371.750000        0.000000  \nmax    11162.000000        1.000000  ",
            "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>AGE</th>\n      <th>BALANCE</th>\n      <th>DURATION</th>\n      <th>PDAYS</th>\n      <th>PREVIOUS</th>\n      <th>CUSTOMERID</th>\n      <th>DEFAULT_STATUS</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>count</th>\n      <td>11162.000000</td>\n      <td>11162.000000</td>\n      <td>11162.000000</td>\n      <td>11162.000000</td>\n      <td>11162.000000</td>\n      <td>11162.000000</td>\n      <td>11162.000000</td>\n    </tr>\n    <tr>\n      <th>mean</th>\n      <td>41.231948</td>\n      <td>1528.538524</td>\n      <td>371.993818</td>\n      <td>51.330407</td>\n      <td>0.832557</td>\n      <td>5581.500000</td>\n      <td>0.015051</td>\n    </tr>\n    <tr>\n      <th>std</th>\n      <td>11.913369</td>\n      <td>3225.413326</td>\n      <td>347.128386</td>\n      <td>108.758282</td>\n      <td>2.292007</td>\n      <td>3222.336187</td>\n      <td>0.121761</td>\n    </tr>\n    <tr>\n      <th>min</th>\n      <td>18.000000</td>\n      <td>-6847.000000</td>\n      <td>2.000000</td>\n      <td>-1.000000</td>\n      <td>0.000000</td>\n      <td>1.000000</td>\n      <td>0.000000</td>\n    </tr>\n    <tr>\n      <th>25%</th>\n      <td>32.000000</td>\n      <td>122.000000</td>\n      <td>138.000000</td>\n      <td>-1.000000</td>\n      <td>0.000000</td>\n      <td>2791.250000</td>\n      <td>0.000000</td>\n    </tr>\n    <tr>\n      <th>50%</th>\n      <td>39.000000</td>\n      <td>550.000000</td>\n      <td>255.000000</td>\n      <td>-1.000000</td>\n      <td>0.000000</td>\n      <td>5581.500000</td>\n      <td>0.000000</td>\n    </tr>\n    <tr>\n      <th>75%</th>\n      <td>49.000000</td>\n      <td>1708.000000</td>\n      <td>496.000000</td>\n      <td>20.750000</td>\n      <td>1.000000</td>\n      <td>8371.750000</td>\n      <td>0.000000</td>\n    </tr>\n    <tr>\n      <th>max</th>\n      <td>95.000000</td>\n      <td>81204.000000</td>\n      <td>3881.000000</td>\n      <td>854.000000</td>\n      <td>58.000000</td>\n      <td>11162.000000</td>\n      <td>1.000000</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
          },
          "metadata": {}
        }
      ],
      "execution_count": 148,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        },
        "ms_comments": [
          {
            "threadId": "987d9472-441a-4810-9d8b-bf7996394c6e",
            "text": "Can we add some more simple EDA things?",
            "status": "resolved",
            "user": {
              "name": "Shankar Narayanan Ganapathy",
              "idType": "aad"
            },
            "createdDateUTC": 1660067375479,
            "modifiedDateUTC": 1661759164178,
            "replies": [
              {
                "replyId": "611904a2-8d2c-423e-bcc0-0bc88b2e1346",
                "text": "will add charts and basic eda stuff later today",
                "user": {
                  "name": "Mash Syed",
                  "idType": "aad"
                },
                "createdDateUTC": 1660328605386,
                "modifiedDateUTC": 1660328605386
              },
              {
                "replyId": "e8e4052c-5196-4368-8e3c-6e334f2bc40e",
                "text": "Hey added some EDA stuff, take a look and let me know what you think",
                "user": {
                  "name": "Mash Syed",
                  "idType": "aad"
                },
                "createdDateUTC": 1660511932680,
                "modifiedDateUTC": 1660511932680
              }
            ]
          }
        ],
        "ms_comment_ranges": {
          "987d9472-441a-4810-9d8b-bf7996394c6e": {
            "text": "describe()",
            "start": {
              "line": 1,
              "column": 4
            },
            "end": {
              "line": 1,
              "column": 14
            }
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Check Target Variable For Class Imbalance\r\n",
        "#### Classes are heavily imbalanced with a default rate of only 1.5%."
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df.DEFAULT_STATUS.value_counts()"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 149,
          "data": {
            "text/plain": "0    10994\n1      168\nName: DEFAULT_STATUS, dtype: int64"
          },
          "metadata": {}
        }
      ],
      "execution_count": 149,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "print('Default rate is:', str(round((df[df.DEFAULT_STATUS == 1].CUSTOMERID.count() / df.CUSTOMERID.count())*100, 1)) + '%') "
      ],
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": "Default rate is: 1.5%\n"
        }
      ],
      "execution_count": 150,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Summarize Variables by Default Status\r\n",
        "#### Below we can see that there is a significant difference in the 'BALANCE', 'PDAYS', AND 'PREVIOUS' columns between customers that default and those that dont."
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df.groupby('DEFAULT_STATUS').mean().reset_index()"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 151,
          "data": {
            "text/plain": "   DEFAULT_STATUS        AGE      BALANCE    DURATION      PDAYS  PREVIOUS  \\\n0               0  41.248772  1552.840640  372.412589  51.818174  0.842550   \n1               1  40.130952   -61.803571  344.589286  19.410714  0.178571   \n\n    CUSTOMERID  \n0  5569.350737  \n1  6376.553571  ",
            "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>DEFAULT_STATUS</th>\n      <th>AGE</th>\n      <th>BALANCE</th>\n      <th>DURATION</th>\n      <th>PDAYS</th>\n      <th>PREVIOUS</th>\n      <th>CUSTOMERID</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0</td>\n      <td>41.248772</td>\n      <td>1552.840640</td>\n      <td>372.412589</td>\n      <td>51.818174</td>\n      <td>0.842550</td>\n      <td>5569.350737</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>1</td>\n      <td>40.130952</td>\n      <td>-61.803571</td>\n      <td>344.589286</td>\n      <td>19.410714</td>\n      <td>0.178571</td>\n      <td>6376.553571</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
          },
          "metadata": {}
        }
      ],
      "execution_count": 151,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df.groupby('DEFAULT_STATUS').mean().reset_index().plot(kind='bar', x='DEFAULT_STATUS', y='BALANCE', rot=0, figsize=(12,6))"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 152,
          "data": {
            "text/plain": "<matplotlib.axes._subplots.AxesSubplot at 0x7fd7cedc00d0>"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 864x432 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "execution_count": 152,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df.groupby('DEFAULT_STATUS').mean().reset_index().plot(kind='bar', x='DEFAULT_STATUS', y='PDAYS', rot=0, figsize=(12,6))"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 153,
          "data": {
            "text/plain": "<matplotlib.axes._subplots.AxesSubplot at 0x7fd7ccaf7550>"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 864x432 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "execution_count": 153,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Aggregate The Data By Education, Default Status\r\n",
        "#### Determining a significant different in BALANCE when factoring in EDUCATION."
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df[['EDUCATION', 'DEFAULT_STATUS', 'BALANCE']].groupby(['EDUCATION', 'DEFAULT_STATUS']).mean().reset_index() #.rename(columns={'CUSTOMERID':'COUNT'})"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 154,
          "data": {
            "text/plain": "   EDUCATION  DEFAULT_STATUS      BALANCE\n0    primary               0  1554.713120\n1    primary               1   -84.000000\n2  secondary               0  1316.348535\n3  secondary               1   -10.463415\n4   tertiary               0  1871.866246\n5   tertiary               1  -126.125000\n6    unknown               0  1780.938525\n7    unknown               1  -115.000000",
            "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>EDUCATION</th>\n      <th>DEFAULT_STATUS</th>\n      <th>BALANCE</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>primary</td>\n      <td>0</td>\n      <td>1554.713120</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>primary</td>\n      <td>1</td>\n      <td>-84.000000</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>secondary</td>\n      <td>0</td>\n      <td>1316.348535</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>secondary</td>\n      <td>1</td>\n      <td>-10.463415</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>tertiary</td>\n      <td>0</td>\n      <td>1871.866246</td>\n    </tr>\n    <tr>\n      <th>5</th>\n      <td>tertiary</td>\n      <td>1</td>\n      <td>-126.125000</td>\n    </tr>\n    <tr>\n      <th>6</th>\n      <td>unknown</td>\n      <td>0</td>\n      <td>1780.938525</td>\n    </tr>\n    <tr>\n      <th>7</th>\n      <td>unknown</td>\n      <td>1</td>\n      <td>-115.000000</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
          },
          "metadata": {}
        }
      ],
      "execution_count": 154,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Correlation Matrix\r\n",
        "#### Not a strong correlation between any of the numeric features and the target column ('DEFAULT_STATUS')."
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df.corr()"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 155,
          "data": {
            "text/plain": "                     AGE   BALANCE  DURATION     PDAYS  PREVIOUS  CUSTOMERID  \\\nAGE             1.000000  0.112300  0.000189  0.002774  0.020169   -0.019550   \nBALANCE         0.112300  1.000000  0.022436  0.017411  0.030805   -0.051580   \nDURATION        0.000189  0.022436  1.000000 -0.027392 -0.026716   -0.454749   \nPDAYS           0.002774  0.017411 -0.027392  1.000000  0.507272   -0.072481   \nPREVIOUS        0.020169  0.030805 -0.026716  0.507272  1.000000   -0.069119   \nCUSTOMERID     -0.019550 -0.051580 -0.454749 -0.072481 -0.069119    1.000000   \nDEFAULT_STATUS -0.011425 -0.060954 -0.009760 -0.036282 -0.035273    0.030502   \n\n                DEFAULT_STATUS  \nAGE                  -0.011425  \nBALANCE              -0.060954  \nDURATION             -0.009760  \nPDAYS                -0.036282  \nPREVIOUS             -0.035273  \nCUSTOMERID            0.030502  \nDEFAULT_STATUS        1.000000  ",
            "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>AGE</th>\n      <th>BALANCE</th>\n      <th>DURATION</th>\n      <th>PDAYS</th>\n      <th>PREVIOUS</th>\n      <th>CUSTOMERID</th>\n      <th>DEFAULT_STATUS</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>AGE</th>\n      <td>1.000000</td>\n      <td>0.112300</td>\n      <td>0.000189</td>\n      <td>0.002774</td>\n      <td>0.020169</td>\n      <td>-0.019550</td>\n      <td>-0.011425</td>\n    </tr>\n    <tr>\n      <th>BALANCE</th>\n      <td>0.112300</td>\n      <td>1.000000</td>\n      <td>0.022436</td>\n      <td>0.017411</td>\n      <td>0.030805</td>\n      <td>-0.051580</td>\n      <td>-0.060954</td>\n    </tr>\n    <tr>\n      <th>DURATION</th>\n      <td>0.000189</td>\n      <td>0.022436</td>\n      <td>1.000000</td>\n      <td>-0.027392</td>\n      <td>-0.026716</td>\n      <td>-0.454749</td>\n      <td>-0.009760</td>\n    </tr>\n    <tr>\n      <th>PDAYS</th>\n      <td>0.002774</td>\n      <td>0.017411</td>\n      <td>-0.027392</td>\n      <td>1.000000</td>\n      <td>0.507272</td>\n      <td>-0.072481</td>\n      <td>-0.036282</td>\n    </tr>\n    <tr>\n      <th>PREVIOUS</th>\n      <td>0.020169</td>\n      <td>0.030805</td>\n      <td>-0.026716</td>\n      <td>0.507272</td>\n      <td>1.000000</td>\n      <td>-0.069119</td>\n      <td>-0.035273</td>\n    </tr>\n    <tr>\n      <th>CUSTOMERID</th>\n      <td>-0.019550</td>\n      <td>-0.051580</td>\n      <td>-0.454749</td>\n      <td>-0.072481</td>\n      <td>-0.069119</td>\n      <td>1.000000</td>\n      <td>0.030502</td>\n    </tr>\n    <tr>\n      <th>DEFAULT_STATUS</th>\n      <td>-0.011425</td>\n      <td>-0.060954</td>\n      <td>-0.009760</td>\n      <td>-0.036282</td>\n      <td>-0.035273</td>\n      <td>0.030502</td>\n      <td>1.000000</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
          },
          "metadata": {}
        }
      ],
      "execution_count": 155,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Plot Histogram\r\n",
        "#### By plotting overlapping histograms of customer account balances, we can see the distribution of values for the account balance variable for customers that default vs. those that dont. It's clear that customers that dont default have most or all of their values that are greater than 0 while customers that do default show a significant."
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df_hist_1 = df[df.DEFAULT_STATUS == 1].BALANCE\r\n",
        "df_hist_0 = df[df.DEFAULT_STATUS == 0].BALANCE.sample(200)\r\n",
        "plt.figure(figsize=(16,6))\r\n",
        "plt.hist(df_hist_0, bins=40, alpha=0.3, label=\"No Default\")\r\n",
        "plt.hist(df_hist_1, bins=40, alpha=0.5, label=\"Default\")\r\n",
        "plt.xlabel(\"Balance\", size=14)\r\n",
        "plt.ylabel(\"Count\", size=14)\r\n",
        "plt.title(\"Distribution of Customer Account Balances\")\r\n",
        "plt.legend(loc='upper right')"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 156,
          "data": {
            "text/plain": "<matplotlib.legend.Legend at 0x7fd7cea41b50>"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 1152x432 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "execution_count": 156,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Feature Engineering\r\n",
        "#### Feature Engineering is the process of transforming your raw data into something more valuable that can help the algorithm predict your outcome. In our dataset, this means two things:\r\n",
        "- Turn categorical columns into binary using the get_dummies function in pandas\r\n",
        "- Rescale numeric columns using the MinMaxScaler function"
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "## create function to tranform numeric and categorical features\r\n",
        "## provide the dataframe and the name of the target column\r\n",
        "def create_features(df, target_col_name, id_col_name): \r\n",
        "    # remove the target col before transformation\r\n",
        "    df_temp = df.loc[:, ~df.columns.isin([target_col_name, id_col_name])]\r\n",
        "    ## For numerics we use min-max rescaling, for categorical we use get_dummies in pandas\r\n",
        "    scaler = MinMaxScaler()\r\n",
        "    ## Here we chain multiple functions together, we transform both numeric and categorial features and join back the target column\r\n",
        "    df_transformed = pd.DataFrame(scaler.fit_transform(df_temp.select_dtypes(include='number')), \r\n",
        "                                        columns=df_temp.select_dtypes(include='number').columns.tolist()).join(pd.get_dummies(df_temp.select_dtypes(include='object'))).join(df[[target_col_name, id_col_name]])\r\n",
        "    \r\n",
        "    df_transformed.columns = [x.upper() for x in df_transformed.columns]\r\n",
        "    return df_transformed"
      ],
      "outputs": [],
      "execution_count": 157,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "df_transformed = create_features(df, 'DEFAULT_STATUS', 'CUSTOMERID')\r\n",
        "df_transformed.head()"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 158,
          "data": {
            "text/plain": "        AGE   BALANCE  DURATION  PDAYS  PREVIOUS  JOB_ADMIN.  JOB_BLUE-COLLAR  \\\n0  0.532468  0.104371  0.268110    0.0       0.0           1                0   \n1  0.493506  0.078273  0.377675    0.0       0.0           1                0   \n2  0.298701  0.092185  0.357566    0.0       0.0           0                0   \n3  0.480519  0.105882  0.148750    0.0       0.0           0                0   \n4  0.467532  0.079851  0.172983    0.0       0.0           1                0   \n\n   JOB_ENTREPRENEUR  JOB_HOUSEMAID  JOB_MANAGEMENT  ...  CAMPAIGN_8  \\\n0                 0              0               0  ...           0   \n1                 0              0               0  ...           0   \n2                 0              0               0  ...           0   \n3                 0              0               0  ...           0   \n4                 0              0               0  ...           0   \n\n   CAMPAIGN_9  POUTCOME_FAILURE  POUTCOME_OTHER  POUTCOME_SUCCESS  \\\n0           0                 0               0                 0   \n1           0                 0               0                 0   \n2           0                 0               0                 0   \n3           0                 0               0                 0   \n4           0                 0               0                 0   \n\n   POUTCOME_UNKNOWN  DEPOSIT_NO  DEPOSIT_YES  DEFAULT_STATUS  CUSTOMERID  \n0                 1           0            1               0           1  \n1                 1           0            1               0           2  \n2                 1           0            1               0           3  \n3                 1           0            1               0           4  \n4                 1           0            1               0           5  \n\n[5 rows x 118 columns]",
            "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>AGE</th>\n      <th>BALANCE</th>\n      <th>DURATION</th>\n      <th>PDAYS</th>\n      <th>PREVIOUS</th>\n      <th>JOB_ADMIN.</th>\n      <th>JOB_BLUE-COLLAR</th>\n      <th>JOB_ENTREPRENEUR</th>\n      <th>JOB_HOUSEMAID</th>\n      <th>JOB_MANAGEMENT</th>\n      <th>...</th>\n      <th>CAMPAIGN_8</th>\n      <th>CAMPAIGN_9</th>\n      <th>POUTCOME_FAILURE</th>\n      <th>POUTCOME_OTHER</th>\n      <th>POUTCOME_SUCCESS</th>\n      <th>POUTCOME_UNKNOWN</th>\n      <th>DEPOSIT_NO</th>\n      <th>DEPOSIT_YES</th>\n      <th>DEFAULT_STATUS</th>\n      <th>CUSTOMERID</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0.532468</td>\n      <td>0.104371</td>\n      <td>0.268110</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>0.493506</td>\n      <td>0.078273</td>\n      <td>0.377675</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>0.298701</td>\n      <td>0.092185</td>\n      <td>0.357566</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>3</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>0.480519</td>\n      <td>0.105882</td>\n      <td>0.148750</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>4</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>0.467532</td>\n      <td>0.079851</td>\n      <td>0.172983</td>\n      <td>0.0</td>\n      <td>0.0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>5</td>\n    </tr>\n  </tbody>\n</table>\n<p>5 rows × 118 columns</p>\n</div>"
          },
          "metadata": {}
        }
      ],
      "execution_count": 158,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Write features back to Snowflake For Batch Inference\r\n",
        "#### We use SQLAlchemy since it doesn't require a pre-defined schema. In our transformed dataframe, we go from 16 columns to 118 columns\r\n",
        "\r\n",
        "#### References: \r\n",
        "- https://docs.snowflake.com/en/user-guide/sqlalchemy.html\r\n",
        "- https://stephenallwright.com/python-connector-write-pandas-snowflake/\r\n",
        "- https://stephenallwright.com/write-to-snowflake-from-pandas-dataframe/"
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "engine = create_engine(URL(\n",
        "    account=sfaccount + '.west-us-2.azure',\n",
        "    user=sfuser,\n",
        "    password=sfpasswd,\n",
        "    role='ACCOUNTADMIN',\n",
        "    warehouse='COMPUTE_WH',\n",
        "    database='DEMO',      \n",
        "    schema ='SOURCE_DATA',\n",
        "    autocommit=True\n",
        "    ))\n",
        "\n",
        "tbl_name = 'banking_customers_features'\n",
        "\n",
        "## drop table if it exists since replace doesnt work in sqlalchemy\n",
        "# Execute a statement that will generate a result set.\n",
        "sql = \"drop table if exists \" + tbl_name\n",
        "cur.execute(sql)\n",
        "\n",
        "df_transformed.to_sql(con = engine, \n",
        "                    name = tbl_name, \n",
        "                    if_exists='replace', #other possible values are 'append' and 'fail',\n",
        "                    schema='SOURCE_DATA',\n",
        "                    index=False)"
      ],
      "outputs": [],
      "execution_count": 162,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Split Data Into Train and Test Sets"
      ],
      "metadata": {
        "application/vnd.databricks.v1+cell": {
          "title": "",
          "showTitle": false,
          "inputWidgets": {},
          "nuid": "a9790b14-49be-4050-b211-402ce198c7f7"
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "X = np.array(np.array(df_transformed.loc[:, df_transformed.columns != 'DEFAULT_STATUS']))\r\n",
        "y = np.array(df_transformed['DEFAULT_STATUS'])\r\n",
        "\r\n",
        "# train/test split\r\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.10, random_state=42)"
      ],
      "outputs": [],
      "execution_count": 163,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Fit the Transformed Features To a Model"
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "## instantiate a base model with no tuning to hyperparameters\r\n",
        "model = RandomForestClassifier(random_state=0, n_jobs=-1)\r\n",
        "print(model)"
      ],
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": "RandomForestClassifier(bootstrap=True, ccp_alpha=0.0, class_weight=None,\n                       criterion='gini', max_depth=None, max_features='auto',\n                       max_leaf_nodes=None, max_samples=None,\n                       min_impurity_decrease=0.0, min_impurity_split=None,\n                       min_samples_leaf=1, min_samples_split=2,\n                       min_weight_fraction_leaf=0.0, n_estimators=100,\n                       n_jobs=-1, oob_score=False, random_state=0, verbose=0,\n                       warm_start=False)\n"
        }
      ],
      "execution_count": 164,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# train the model on the training data set\r\n",
        "model.fit(X_train, y_train)"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 165,
          "data": {
            "text/plain": "RandomForestClassifier(bootstrap=True, ccp_alpha=0.0, class_weight=None,\n                       criterion='gini', max_depth=None, max_features='auto',\n                       max_leaf_nodes=None, max_samples=None,\n                       min_impurity_decrease=0.0, min_impurity_split=None,\n                       min_samples_leaf=1, min_samples_split=2,\n                       min_weight_fraction_leaf=0.0, n_estimators=100,\n                       n_jobs=-1, oob_score=False, random_state=0, verbose=0,\n                       warm_start=False)"
          },
          "metadata": {}
        }
      ],
      "execution_count": 165,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Score The Model Using 5-Fold Cross Validation"
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "## score the model\r\n",
        "scores = cross_val_score(model, X_train, y_train, cv=5)\r\n",
        "print(\"Mean cross-validation score: %.2f\" % scores.mean())"
      ],
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": "Mean cross-validation score: 0.99\n"
        }
      ],
      "execution_count": 166,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Test the Fitted Model On Unseen Data\r\n",
        "#### Now that we have trained our model and run cross validation on our training data, we can apply our model to unseen observations which will allow us to understand how well it performs"
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "y_pred = model.predict(X_test)"
      ],
      "outputs": [],
      "execution_count": 167,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Use the Confusion Matrix To Understand Accuracy\r\n",
        "#### Earlier in the notebook, we discussed class imbalance. The confusion matrix shows us how the skew of the default variable impacts our accuracy. There are ways to improve class imbalance but that is outside of the scope of this use case."
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "## use the confusion matrix to understand accuracy\r\n",
        "cm = confusion_matrix(y_test, y_pred)\r\n",
        "TN, FP, FN, TP = confusion_matrix(y_test, y_pred).ravel()\r\n",
        "print('True Positive(TP)  = ', TP)\r\n",
        "print('False Positive(FP) = ', FP)\r\n",
        "print('True Negative(TN)  = ', TN)\r\n",
        "print('False Negative(FN) = ', FN)\r\n",
        "\r\n",
        "accuracy =  (TP+TN) /(TP+FP+TN+FN)\r\n",
        "print('Accuracy of the classification = {:0.3f}'.format(accuracy))"
      ],
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": "True Positive(TP)  =  0\nFalse Positive(FP) =  0\nTrue Negative(TN)  =  1092\nFalse Negative(FN) =  25\nAccuracy of the classification = 0.978\n"
        }
      ],
      "execution_count": 168,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Deploy The Trained Model To Snowpark\r\n",
        "#### Open the connection to Snowflake and create a new Snowpark session\r\n",
        "#### Reference : https://docs.snowflake.com/en/developer-guide/snowpark/python/index.html\r\n",
        "\r\n"
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "connection_parameters = {\r\n",
        "  \"account\": sfaccount + '.west-us-2.azure',\r\n",
        "  \"user\": sfuser,\r\n",
        "  \"password\": sfpasswd,\r\n",
        "  \"role\": 'ACCOUNTADMIN',\r\n",
        "  \"warehouse\": 'COMPUTE_WH',\r\n",
        "  \"database\": 'DEMO',\r\n",
        "  \"schema\": 'SOURCE_DATA'\r\n",
        "}\r\n",
        "\r\n",
        "session = Session.builder.configs(connection_parameters).create()\r\n",
        "df = session.create_dataframe([[1, 2], [3, 4]], schema=[\"a\", \"b\"])\r\n",
        "df = df.filter(df.a > 1)\r\n",
        "df.show()\r\n",
        "pandas_df = df.to_pandas()  # this requires pandas installed in the Python environment\r\n",
        "result = df.collect()"
      ],
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": "-------------\n|\"A\"  |\"B\"  |\n-------------\n|3    |4    |\n-------------\n\n"
        }
      ],
      "execution_count": 170,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Create a Stage To Store the UDF and libraries required for deployment "
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#Let's first create a stage to store the artifacts\r\n",
        "print(session.sql('create stage if not exists MODELSTAGE').collect())"
      ],
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": "[Row(status='MODELSTAGE already exists, statement succeeded.')]\n"
        }
      ],
      "execution_count": 171,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "session.add_packages(\"pandas\", \"numpy\", \"scikit-learn==1.0.2\")"
      ],
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": "WARNING:root:The version of package scikit-learn in the local environment is 0.22.1, which does not fit the criteria for the requirement scikit-learn==1.0.2. Your UDF might not work when the package version is different between the server and your local environment\n"
        }
      ],
      "execution_count": 172,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Register your trained model as a UDF in your instance of Snowflake"
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "features = df_transformed.columns[:-2].tolist()\r\n",
        "predict_churn_model = session.udf.register(lambda *args: \r\n",
        "                                    model.predict(pd.DataFrame(args, columns=features)),\r\n",
        "                                    name=\"predict_churn_model\",\r\n",
        "                                    stage_location=\"@MODELSTAGE\",\r\n",
        "                                    return_type=FloatType(),\r\n",
        "                                    is_permanent=True,\r\n",
        "                                    replace=True,\r\n",
        "                                    input_types=[ArrayType()])"
      ],
      "outputs": [],
      "execution_count": 173,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "@udf(name='predict_default',is_permanent = True, stage_location = '@MODELSTAGE', replace=True, session=session)\r\n",
        "def predict_default(args: list) -> float:\r\n",
        "    row = pd.DataFrame([args], columns=features)\r\n",
        "    return model.predict(row)"
      ],
      "outputs": [],
      "execution_count": 174,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Apply the deployed model on a random set of customer ID's\r\n",
        "#### Test the deployed model (batch inference) out on a random sample of records and write the results to a new table. In this case, we use the feature table (as input) we created in a previous step."
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "default_prediction_df = session.table('BANKING_CUSTOMERS_FEATURES').sample(n = 400)\r\n",
        "default_prediction_df.select(default_prediction_df.CUSTOMERID, default_prediction_df.DEFAULT_STATUS, \\\r\n",
        "              F.call_udf(\"predict_default\", F.array_construct(*features)).alias('PREDICTED_DEFAULT')) \\\r\n",
        "        .write.mode('overwrite').saveAsTable('default_detection')"
      ],
      "outputs": [],
      "execution_count": 175,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Final Table with Predicted Values\r\n",
        "#### Connect to the new table with predictions and show in a pandas dataframe."
      ],
      "metadata": {
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Execute a statement that will generate a result set.\r\n",
        "sql = \"select * from demo.source_data.default_detection\"\r\n",
        "cur.execute(sql)\r\n",
        "\r\n",
        "# Fetch the result set from the cursor and deliver it as the Pandas DataFrame.\r\n",
        "df = cur.fetch_pandas_all()\r\n",
        "\r\n",
        "df.head()"
      ],
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 176,
          "data": {
            "text/plain": "   CUSTOMERID  DEFAULT_STATUS  PREDICTED_DEFAULT\n0        3167               0                0.0\n1       10634               0                0.0\n2        6504               0                0.0\n3        9544               0                0.0\n4        9566               0                0.0",
            "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>CUSTOMERID</th>\n      <th>DEFAULT_STATUS</th>\n      <th>PREDICTED_DEFAULT</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>3167</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>10634</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>6504</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>9544</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>9566</td>\n      <td>0</td>\n      <td>0.0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
          },
          "metadata": {}
        }
      ],
      "execution_count": 176,
      "metadata": {
        "jupyter": {
          "source_hidden": false,
          "outputs_hidden": false
        },
        "nteract": {
          "transient": {
            "deleting": false
          }
        }
      }
    }
  ],
  "metadata": {
    "application/vnd.databricks.v1+notebook": {
      "notebookName": "Snowflake Demo Project (Abalone Model) v2",
      "dashboards": [],
      "notebookMetadata": {
        "pythonIndentUnit": 2
      },
      "language": "python",
      "widgets": {},
      "notebookOrigID": 3281562606889750
    },
    "language_info": {
      "name": "python",
      "version": "3.8.5",
      "mimetype": "text/x-python",
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "pygments_lexer": "ipython3",
      "nbconvert_exporter": "python",
      "file_extension": ".py"
    },
    "kernelspec": {
      "name": "python38-azureml",
      "language": "python",
      "display_name": "Python 3.8 - AzureML"
    },
    "kernel_info": {
      "name": "python38-azureml"
    },
    "nteract": {
      "version": "nteract-front-end@1.0.0"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}