Seynro/Credit_Scoring_Model

A database of 33,500 micro and small business loans disbursed between July 2020 and March 2022 by a fictitious bank in Kenya. The loans can be denominated in local currency or in USD, but all values in the database are represented in USD equivalent. # ML Task Overview ## Data Description: - A database of 33,500 micro and small business loans disbursed between July 2020 and March 2022 by a fictitious bank in Kenya. - The loans can be denominated in local currency or in USD, but all values in the database are represented in USD equivalent. ## Work Completed: 1. **Data Preprocessing**: Cleaned the data and handled missing values. Converted the categorical variables into binary modalities for both the bonus task and the scoring model. 2. **Feature Selection**: Excluded contract-related variables such as disbursement date, loan amount, maturity, and effective annual rate. Utilized domain knowledge and correlation analysis for other features. 3. **Model Development**: Developed a logistic regression model. Segregated approximately 20% of the observations for "out-of-sample" validation. 4. **Performance Evaluation**: Evaluated the model's performance using both primary and validation classification tables. 5. **Recommendation**: Provided strategic insights based on the scoring model's outcomes and suggested implementation steps for the credit department.