Credit-Worthiness-in-Microfinance-Banks-using-Explainable-AI

While giving credit scores to applicants on their loan applications, here are the following issues that can be faced:

1. Bankers can have trust issues over Machine Learning models and they can find those models intransparent and unreliable. To address transparency issues in credit worthiness, this project uses Explainable AI to bring interpretability in Machine Learning models.

2. As bank loan datasets are very imbalanced because more applications are not credit worthy, so to address this data imbalance, I've used Synthetic Minority Over-Sampling Technique (SMOTE) to ensure that the data is balanced. This improves the evaluation metrics of Machine Learning Models and make the models unbiased towards both the parties i.e,. Credity Worhty and Non-Credit Worthy Parties.

Here is the link to view code on Google Colab Notebook: https://colab.research.google.com/drive/1neBiGDRewDBmD8PK1fz2cXi6wG19iUkK#scrollTo=NomW4rJnEmVw