Machine Learning project to predict credit default based on the customer's monthly profile.
- The dataset is from the American Express - Default Prediction kaggle competition.
- Goal of the competition is to predict credit default based on the customer's monthly profile.
- Raw data
- Simplelmputer
- Simplelmputer
- OrdinalEncoder
- OneHotEncoder
- RemoveMulticollinearity
- FixImbalancer
- StandardScaler
- SelectFromModel
- CatBoostClassifier
Git LFS is used to store the orginal dataset. The CSV file was converted to the feather format to reduce the size of the file.
MLflow is used to track model training, model performance, and model deployment.
The project will be deployed using a combination of Pycaret and the Gradio library.
- Pycaret will be used for the data preprocessing, model training, and model selection.
- Gradio will be used to create a web app that will allow users to interact with the model.
- FastAPI will be used to create an API that will allow the web app to communicate with the model.
- The web app will be hosted on Azure.