Credit Card Approval System

Duration - 16.02.2020 to 30.06.2020

Design Lab Project . Watch the technical demo here

Deployment

Deployed using Github Actions. Check this build for example

Team

• Muskan - Project Manager
• Sayantan Das - Lead Developer
• Sheetal Kumari - QA Manager
• Soujannya Roy - QA Engineer
• Md Salman - Developer
• Debankan Ganguly - Developer

Problem Statement

Small businesses must seek credit approval to obtain funds from lenders, investors, and vendors, and also grant credit approval to their customers. Banking industries receive so many applications for credit card request. Going through each request manually can be very time consuming, also prone to human errors.

Solution Statement

• I : Building a probabilistic statistical model that looks into historical data to find patterns in what parameters lie in decision making behind the approval of a credit model.
• II : Deploying a trained model.The bank server makes a FTP request to Maven , sending it their list of applications for the day. Maven's AI cron job does the predictions and sends it back to the bank.
• III : To ensure model robustness, model is refreshed every 5 days by developers working on Maven. They employ public/privatised data and a myriad of boosted trees and other advanced algorithms to make sure misclassification can be avoided as much as possible during inference phase.

Initial Setup

sh setup.sh

If you are running on Anaconda follow these instructions,

conda create -n mavenlab python=3.6
conda activate mavenlab

Quick Run

 cd src/
 sh run.sh

Tests

cd tests/
sh runtest.sh

FTP instructions are posted here

Model Creation

Preconfigured Notebook can be viewed here.

Use Case Diagram

Class Diagram

Sequence Diagrams

WBS