Note: The application was deployed using free services and hence is not available starting from 30th October 2023. Please get in touch with me in order to see a working demo.
This repository contains all the code for training, testing and deploying a predictive model for "Prêt à dépenser" in order to predict if a client can or cannot repay his loan back with difficulties and it's probability. A data drift report generated by Evidently can also be found in this repo alongside notebooks for exploratory analysis, training, feature engineering and hyper parameter optimization.
The API along with the model for inference can be found here.
The interactive dashboard for this application is hosted here Click on it to access the dashboard. Here are a few client IDs to test the dashboard with:
- 100005
- 100013
- 100067
- 100066
Note: The client information is editable so, feel free to play around by changing the client's income (for example) and see how it affects the decision taken by the model.