Introduction

Welcome to our workshop! In this workshop we'll be using the Cloud Pak for Data platform to Collect Data, Organize Data, Analyze Data, and Infuse AI into our applications. The goals of this workshop are:

• Visualize data with Data Refinery

• Create and deploy a machine learning model

• Monitor the model

• Create a Python app to use the model

About this workshop

• Agenda

• Compatability

• About Cloud Pak for Data

• Credits

About the data set

In this workshop we will be using a credit risk / lending scenario. In this scenario, lenders respond to an increased pressure to expand lending to larger and more diverse audiences, by using different approaches to risk modeling. This means going beyond traditional credit data sources to alternative credit sources (i.e. mobile phone plan payment histories, education, etc), which may introduce risk of bias or other unexpected correlations.

Use Case Diagram

The credit risk model that we are exploring in this workshop uses a training data set that contains 20 attributes about each loan applicant. The scenario and model use synthetic data based on the [UCI German Credit dataset](https://archive.ics.uci.edu/ml/datasets/Statlog+(German+Credit+Data)). The data is split into three CSV files and are located in the data directory of the GitHub repository you will download in the pre-work section.

Applicant Financial Data

This file has the following attributes:

• CUSTOMERID (hex number, used as Primary Key)

• CHECKINGSTATUS

• CREDITHISTORY

• EXISTINGSAVINGS

• INSTALLMENTPLANS

• EXISTINGCREDITSCOUNT

Applicant Loan Data

This file has the following attributes:

• CUSTOMERID

• LOANDURATION

• LOANPURPOSE

• LOANAMOUNT

• INSTALLMENTPERCENT

• OTHERSONLOAN

• RISK

Applicant Personal Data

This file has the following attributes:

• CUSTOMERID

• EMPLOYMENTDURATION

• SEX

• CURRENTRESIDENCEDURATION

• OWNSPROPERTY

• AGE

• HOUSING

• JOB

• DEPENDENTS

• TELEPHONE

• FOREIGNWORKER

• FIRSTNAME

• LASTNAME

• EMAIL

• STREETADDRESS

• CITY

• STATE

• POSTALCODE

Agenda

Pre-work	Creating a project, downloading the data set, seeding a database
Data Visualization with Data Refinery	Refining the data, vizualizing and profiling the data
Machine Learning with Jupyter	Building a model with Spark, deploying the model with Watson Maching Learning, testing the model with a Python Flask app
Machine Learning with AutoAI	Use AutoAi to quickly generate a Machine Learning pipeline and model
Deploy and Test Machine Learning Models	Deploy and machine learning models using several approaches

Compatability

This workshop has been tested on the following platforms:

• macOS: Mojave (10.14), Catalina (10.15)

  • Google Chrome version 81

• Microsoft: Windows 10

  • Google Chrome, Microsoft Edge