Cloud Pak for Data Group

SME Banking 

Wed June 17, 2020 12:40 PM

Making Small/Medium Business Banking Easier

The SME Banking accelerator demonstrates a model that analyzes existing known financial factors of a small business, and automatically pre-approves a loan. Further, upon request for a larger amount loan amount, the model evaluates the risk and recommends further actions needed to acquire the loan.

Introduction

Evaluating a loan risk is challenging and costly for SME banking institutions. The SME Banking accelerator uses machine learning and explainable AI solutions to analyze existing financial information and known factors of the business. The model evaluates the risk of a requested loan, and automatically approves or rejects it for a customer. Further, upon request for an approved loan, the model recommends a loan amount to the customer. Lending institutions and banks can use our accelerator to reduce costs and risks associated to loans and make better decisions faster.

The SME Banking Prediction accelerator includes a set of sample data science assets. The Data assets provides the information that you need to predict loan status, such as acceptance of loan, life of loan, and amount of loan. Your data scientists can use the sample notebooks, predictive models, and UI provided to accelerate data preparation, machine learning modeling, and data reporting.

Summary

Context

Financial institutions face a number of challenges throughout the Small / Medium Enterprise lending value chain that are preventing them from addressing the current gap in lending. These challenges include:

  • Minimal or poor integration of SME data with credit bureaus and other financial tools.
  • Lack of proper process auditability because of applications’ and approvals’ manual processes.
  • Data Collection, payment, and collections is a manual process, thus difficult to track.

Solution

Machine learning and explainable AI solutions to evaluate the risks and to facilitate decision-making.

  • Acceptance classifier predicts the most likely outcomes of an applied loan (Accept/Reject).
  • Loan life classifier predicts the life of a loan (longer/shorter than one year).
  • Credit limit regressor suggests an optimal amount of that yields the maximum profit for the lending agency.
  • Interest rate regressor informs the customer about an estimation of the loan interest rate.

Accelerator use case

The accelerator will provide AI solutions to help lending agencies process loan data more efficiently. This process can be improved by equipping SME customer with relevant AI predictions.

The accelerator will provide a sample application that the SME customer can use. The sample application will have the following:

  • Prediction of a pre-approved loan amount based on the customer historical data.
  • Additional information such as estimated interest rate and monthly payment become available as soon as the customer inputs data for a new loan.
  • Prediction of the most likely outcome of the loan status (Accept/Reject) if the customer applies for a new loan.
  • Predicted amount of the new loan that is recommended for a customer based on the updated information.

An amount will be recommended only if the loan is accepted and the life of loan is predicted as longer than one year. The model considers a wide range of values based on the customer's applied amount and recommends the amount that yields the maximum profit.

User groups for CP4D services

  • The analytics team will have access to a sample data preparation notebook. The scripts give an idea of how they can prepare their organization data into a feature set for training models.
  • The analytics team will have access to machine learning models built by AutoAI, which they can run by clicking a few buttons. There is also a Jupyter notebook on predicting loan status, life, and amount, for advanced data scientists in the team. Also, key factors that influenced the prediction of the most likely outcome, are visible from AutoAI and notebooks.
  • The development team will see how they can easily deploy machine learning models with Watson Machine Learning.
  • The SME customer will be able to use a custom user interface to key in details and get the AI predictions.

Sample Application

This is Natalie Smith, the owner of Watson Café who would like to apply for a loan. A sample loan application is provided for Natalie to interact with the deployed models.

This is how Natalie, the loan applicant, would interact with the application.

1. Natalie is the SME customer who wants to apply for a loan. She first looks at her checking and saving accounts information. If she is eligible for a loan, she can view a pre-approved amount for the loan that is recommended based on her past information. She can apply for a new amount and click on 'Start Application' to start the loan application process.

2. Natalie keys in details of her business including the term and the amount of the loan and views the estimated interest rate and monthly payment which are calculated based on her input. She clicks on the 'Continue with application' button.

3. Natalie might need to upload additional documents if she needs to update her information and convince the lending agency for a higher loan amount.

4. Natalie sees a new screen with the decision on the loan application. This page includes the following information.

  • Top panel: Information about the loan status and whether the amount of loan requested is accepted.
  • Bottom left panel: If the loan is accepted, a new loan amount is suggested immediately instead of sending her through a long process. Here, additional information such as the term of the loan is provided. Natalie can accept the loan here.
  • Bottom right panel: Natalie can choose to call an SME advisor.

Installing and running the accelerator

Prerequisites

Required services: This accelerator requires IBM Cloud Pak for Data base component installed.

Importing the accelerator

To use this accelerator on Cloud Pak for Data v3.0.1, contact the Data Science Elite team.

Release Notes

This accelerator has been verified on:

      • Cloud Pak for Data v3.0.1

About the developer:

IBM

Terms and Conditions

The terms under which you are licensing IBM Cloud Pak for Data also apply to your use of the Industry Accelerators. Before you use the Industry Accelerators, you must agree on these additional terms and conditions that are set forth here. This information contains sample modules, exercises, and code samples (the code may be provided in source code form ("Source Code")) (collectively "Sample Materials").

License: Subject to the terms herein, you may copy, modify, and distribute these Sample Materials within your enterprise only, for your internal use only; provided such use is within the limits of the license rights of the IBM agreement under which you are licensing IBM Cloud Pak for Data. The Industry Accelerators might include applicable third-party licenses. Review the third-party licenses before you use any of the Industry Accelerators. You can find the third-party licenses that apply to each Sample Material in the notices.txt file that is included with each Sample Material.

Code Security: Source Code may not be disclosed to any third parties for any reason without IBM's prior written consent, and access must be limited to your employees who have a need to know. You have implemented and will maintain the technical and personnel focused security policies, procedures, and controls that are necessary to protect the Source Code against loss, alteration, unlawful forms of processing, unauthorized disclosure, and unauthorized access. You will promptly (and in no event any later than 48 hours) notify IBM after becoming aware of any breach or other security incident that you know, or should reasonably suspect, affects or will affect the Source Code or IBM, and will provide IBM with reasonably requested information about such security incident and the status of any remediation and restoration activities. You will not permit any Source Code to reside on servers located in the Russian Federation, the People's Republic of China, or any territories worldwide in which the Russian Federation or People's Republic of China claim sovereignty (collectively, "China or Russia"). Company shall not permit anyone to access or use any such Source Code from or within China or Russia, and Company will not permit any development, testing, or other work to occur in China or Russia that would require such access or use. Upon reasonable written notice, IBM may extend these restrictions to other countries that the United States government identifies as potential cyber security concerns. IBM may request that you verify compliance with these Code Security obligations, and you agree to cooperate with IBM in that regard.

General: Notwithstanding anything to the contrary, IBM PROVIDES THE SAMPLE MATERIALS ON AN "AS IS" BASIS AND IBM DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, ANY IMPLIED WARRANTIES OR CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, FITNESS FOR A PARTICULAR PURPOSE, TITLE, AND ANY WARRANTY OR CONDITION OF NON-INFRINGEMENT. IBM SHALL NOT BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY OR ECONOMIC CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR OPERATION OF THE SAMPLE MATERIALS. IBM SHALL NOT BE LIABLE FOR LOSS OF, OR DAMAGE TO, DATA, OR FOR LOST PROFITS, BUSINESS REVENUE, GOODWILL, OR ANTICIPATED SAVINGS. IBM HAS NO OBLIGATION TO PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS OR MODIFICATIONS TO THE SAMPLE MATERIALS.


#Featured-area-2
#datascience
#CloudPak4Data
#cloud-pak-for-data
#cloudpakfordata
#Featured-area-3

Statistics

0 Favorited
87 Views
0 Files
0 Shares
0 Downloads