Classification model for in-vehicle coupon acceptances.
This project aims to explore relationships between the data in the UCI in-vehicle coupon dataset and use gained insights to train a model that can accurately predict whether the driver will accept the coupon. The project incorporates initial inspection of the data, exploratory data analysis via visualizations and groupby operations, data cleaning and processing, and model training and evaluation to accomplish these goals.
The maximal F1-statistic for the correct prediction of coupon acceptances was 0.80 - 0.81. This value varied due to the variability among iterations of training and testing with the same model.
Tools used: Python, NumPy, Pandas, Matplotlib, Seaborn, Scikit-learn, XGBoost, Pickle.