Welcome to my repository for the credit risk classification. Please explore the codebase!

This project presents a classification model that identifies creditworthiness of borrowers.

This project leverages python 3.7 with the following packages:

• pandas - Library for reading/writing csv files and fast manipulation with DataFrames.

• numpy - Library for working with arrays.

• pathlib - Library for filesystem paths.

• sklearn - Library for predictive-data analysis.

• imblearn - Library for working with imbalanced classes.

Before running the application first install the following dependencies.

from imblearn.metrics import classification_report_imbalanced
import warnings

  pip install pandas
  pip install numpy
  pip install pathlib
  pip install -U scikit-learn
  pip install imblearn

To use this project simply clone the repository and run the code credit_risk_resampling.ipynb in JupyterLab or in VS Code.

This script uses classification model, logistic regression model, to predict creditworthiness of borrowers. The algorithm overcomes the problem with imbalanced classes by resampling the training set. The dataset takes on the following form:

The first model runs on the randomly spit training/testing datasets without resampling.

The accuracy measures indicate quite a high forecasting accuracy for this first model:

The second model runs on the resampled dataset:

and also proves to be very accurate:

MIT