ELO MERCHANT CATEGORY RECOMMENDATION COMPETITION

PURPOSE: To help understanding customer loyalty. This machine learning competition was hosted by Kaggle where we need to develop a regression model that will predict the loyality scores given for each card_id.

Competition: https://www.kaggle.com/c/elo-merchant-category-recommendation

Dataset source: https://www.kaggle.com/c/elo-merchant-category-recommendation/data

REQUIRED PYTHON LIBRARIES: Numpy, Pandas,Matplotlib,Seaborn,Scikit-learn, lightgbm, xgboost, datetime, gc, Optuna, Keras, Flask.

DESCRIPTION: Please execute notebook-elo-prediction.ipynb for complete execution. To check the loyalty score for a particular customer, execute the Final.ipynb and update the parameters as required. In order to run it as a web app in your localhost please execute app.py and access the URL as :Port/index

BEST MODEL: LGBM Regressor with stratified K-fold gives us the lowest RMSE, optuna is used for hyperparameter tuning.

LICENSE: Distributed under the MIT License.

Pull requests are welcome.

Please visit the blog https://raisulhazari.medium.com/elo-predicting-customer-loyalty-score-4e0ddd73c468 for step by step approach and
https://loyaltyscore.herokuapp.com/index to access the deployed web app