The data is taken from https://www.kaggle.com/c/south-german-credit-prediction/overview. The objective is to predict if the customer will default on their credit in the bank. One of the aim of this notebook is not only to achieve high accuracy of prediction, but also to provide an interpretable model as it is important in the financial institution according to the EU regulation, whenever Machine Learning are used for decision making, institution need to be able to have transparency on how machine learning make the decission.
Tools:
- Model: Lazy Classifier, Random Forest Classifier
- Visualization: plotly, seaborn
- Machine Learning Interpretability: SHAP & LIME