README.md

CREDIT_RISK_MODELLING_PROJECT

** Project_Case study:

• Preprocessing, Modeling, Model validation and Maintenance in Python

Business Problem

The task at hand for Instacart is two fold.

1. Customer Profiling - Understanding the different mix of your customers is the key to any successful business. What are the different customer segments? How do their behaviur patterns differ from each other?
2. Customer Targeting - Which segments to target to maximize profitability?
3. Product Recommendations - Once you've identified your target customers, what products to recommend?

Created and Validated PD Model:

Prepared PD Model: Fine classing, Weight of evidence, & coarse classing, Information value, Automating calculations, visualizing results, creating dummies [for both Discrete and Continuous data]

Estimated PD Model: Built Logistic regression model with p-values and interpreted the coefficients in the PD model

Validated PD Model: Out-of-sample validation, Evaluated the model performance(Accuracy and AUC), Gini and Kolmogorov-Smirnov

Applied the PD Model for Decision making: Calculated probability of default for a single customer, Create Scorecard, Calculated Credit score, Setting Cut-offs

Monitored the PD Model: Via Assessing population stability, Index preprocessing, calculated & Interpreted

Worked on LGD and EAD Models: Distribution of recovery rates and credit conversion factors

1. On LGD Model:

• Prepared the inputs, testes the model, Estimated the accuracy, saved the model, Build Linear regression and evaluated.

2. On EAD model:

• Estimated and Interpreted, validated the model, built and updated EAD model

Calculated Expected Loss

Data Science Techniques Used:

• Weight of evidence

• Information value

• Fine classing

• Coarse classing

• Linear regression

• Logistic regression

• Area Under the Curve

• Receiver Operating Characteristic Curve

• Gini Coefficient

• Kolmogorov-Smirnov

• Assessing Population Stability

• Maintaining a model