Table of Contents
- Problem Statement
- Case Study Lifecycle
Problem Statement
Credit risk is defined as the risk of loss resulting from the failure by a borrower to repay the principal and interest owed to the leader. The lender uses the interest payments from the loan to compensate for the risk of potential losses. When the borrower defaults on his/her obligations, it causes an interruption in the cash flows of the lender.
Most of the bank's wealth is obtained from providing credit loans so that a marketing bank must be able to reduce the risk of non-performing credit loans. The risk of providing loans can be minimized by studying patterns from existing lending data. One technique that you can use to solve this problem is to use data mining techniques. Data mining makes it possible to find hidden information from large data sets by way of classification. By using machine learning classification algorithms, we can build a model to predict whether the person, described by the attributes of the dataset, is a good (1) or a bad (0) credit risk.
Case Study Lifecycle:
-
Data Exploration : Started exploring dataset using pandas,numpy,matplotlib and seaborn.
-
Data visualization : Ploted graphs to get insights about dependend and independed variables.
-
Feature Engineering : All The Value Are Arrange In One Range.
-
Model Selection I : Tested all base models to check the base accuracy.
-
Model Selection II : Performed Hyperparameter tuning using gridsearchCV.
-
Pickle File : Selected model as per best accuracy and created pickle file using Pickle .
-
Webpage & deployment : 1. Created a flask web form that takes all the necessary inputs from user and shows output. 2. Deployment at Heroku Platform
Document Link: https://drive.google.com/drive/folders/18pXqwkFdB9tWqo0hYThyQ9c0vVRFp1ZA?usp=sharing