This journey accesses a Financial Risk Management API published on IBM Cloud with Machine Learning on z/OS running on the IBM Z Mainframe through a simulated retail bank system called MPLbank.
This journey accesses a fictitious retail banking system called MPLbank. MPLbank integrates a Financial Risk Management System based on Machine Learning for z/OS. The initial predictive model was designed to deliver a score representing the probability of the capacity of loan refund for a banking customer according to their personal data. On top of these components, an API layer hosted in IBM Cloud has been set up to deliver a Financial Risk Management API, illustrating an online decision for loans approval.
Following the below schema, the predictive model was built from data using IBM Machine learning tools. Banking customer data was selected (age, number of credit card, income and number of car loans) to predict a score (probability for capacity of loan refunding) and an action (next customer's behaviour). It was then trained using machine learning with 80% of historical data and evaluated with 20% of the historical data. This ensures the model is able to learn from fresh data.
Once the model was approved, it was deployed to act as a scoring service. On top of this service, an API was created and published to an API developer Portal and is now callable through API.
The objectives of this journey are to discover, test and use this Financial Risk Management API using a sample financial application, then enhance it using IBM Cloud.
The journey is accomplished by using a Hybrid IBM Cloud / IBM Z Mainframe architecture:
- Start with the API Developer Portal
- Subscribe to the Financial Risk Management API
- Work with the Financial Risk Management API
-
Sign up for an IBM ID if you don't have one already.
-
Go to the API Developer Portal.
-
Create an account if you have not done that already.
- Click Create an Account.
- Provide all required information. Be sure to use your IBM ID (email address) for this account.
- Click Submit.
An account activation email will be sent to your registered IBM ID email. Click on the link in this email to activate your account.
-
Login to your account.
-
- Click Apps from the menu.
- Click Create new App.
- Fill in all required fields.
- Click Submit.
Make a note of the client ID and client Secret. You will need them to access the API later.
-
Before working with the Financial Risk Management API, you need to subscribe to it first. Display the list of available API products.
- Click API Products from the top menu.
- Click Banking Product in the list.
-
- Click Subscribe to the Default Plan.
- Select the App that you have just created before.
- Click Subscribe.
-
- Click Banking APIs.
This page has 3 sections:
- The left panel is the navigation panel that lists all the available operations and their definitions.
- The middle panel displays detail information for the item you have selected.
- The right panel contains sample code in various programming languages.
-
Discover the operation Get /customers/loan/calculateScore by reading its documentation.
- Click Get /customers/loan/calculateScore.
-
Generate code for the operation Get /customers/loan/calculateScore following the right panel of this operation.
- Click a programming language that you want to work with.
Sample code for the selected programming language and an example output of a successful response are displayed. You can copy the code and use it in your own application.
-
Test the operation Get /customers/loan/calculateScore in your programming language with Input parameters:
Parameters Value example Age Integer 23 Income Integer 30000 Number of credit Card Integer 2 Number of car loan Integer 1 -
Scroll down to the Try this operation section.
-
Fill with Input values.
IMPORTANT: All available customers ID are in the /identifier/customerIDs.txt file in this Github repository. Do not forget to fill the x-ibm-client-id and x-ibm-client-secret with yours.
-
Click Call Operation.
You should see the returned output at the bottom of the page.
A score and a message is returned.
-
π Congratulations! You have successfully discovered and tested the Financial Risk Management API.
A quick financial application has been developed in order to help you to start coding. This web application (HTML/CSS/Javascript) uses the Financial Risk Management API introduced before.
-
Download and import the project financialApplication located in this Github repository into you preferred IDE like Eclipse.
- Either click on Download ZIP
- Or use Git Command :
git clone https://github.com/IBM/Financial-risk-management-using-machine-learning-on-zSystems.git
-
Review the index.html file in order to understand how it is working.
-
Review the financialAPI.js file in order to understand how the script works.
- Replace IBM_CLIENT_ID & IBM_CLIENT_SECRET variables by yours and save the file.
-
Open the index.html in your favorite web browser. The application will automatically run.
NOTE: There is no need to compile JS/HTML/CSS from any IDE! Just edit those files in the IDE and refresh the index.html* in the web browser (or Ctrl + F5 shortcut key) to reload this web application.
-
Fill input values with previous values example then Click on the button Go.
This will call the published operation Get /customers/loan/calculateScore.
π Congratulations! You have successfully developed your first financial application.
-
NOTE: Use IBM Cloud to create, test and deploy a quick application. Choose among JAVA Liberty Profile, Node.js servers, Ruby, Python, etc. This platform also provides DevOps tools for a continuous delivery (Git, automatic deployment) and a lot of innovative features & services.
-
Configure your Node.js project for free (30 days).
- Provide an App name.
- Select a domain.
- Select a region to deploy the project.
- Click Create.
-
Wait for the Node.js Runtime creation.
-
Once created, explore this panel to be familiar with it.
- Click Visit App URL. The default Node.js Project (Hello World) has been provided.
- Explore each menu on the left panel to understand the Runtime, connections, logs, etc.
- Read the Getting Started.
-
Clone the project from Git https://github.com/IBM-Cloud/get-started-node.git
-
Edit the cloned Hello World sample application on your laptop to integrate the Financial application files:
- Copy the js folder from the financial application into the views folder.
- Remove the views/stylesheets* folder to delete the default CSS style.
- Copy the css folder from the financial application into the views folder.
- Replace the index.html from the financial application to the views folder.
- Edit the manifest.yml to change the app name and delete the random route line.
-
Re-Deploy the new code to the Node.js Runtime in Cloud using the bx login and bx app push command you read in step 5.
The Financial Risk Management application is now hosted in Cloud and use the Finance Risk Management API.
π Congratulations! You have successfully developed your first financial application in Cloud.