lenta--hackathon

Approach to solving the tape hackathon, 2nd place ( public + private leaderboard)

Input files: hist_data.csv - file with date for training test.csv - a file with a date for validating a solution at a hackathon

Notebooks: eda.ipynb - exploratory analysis user_class.ipynb - classification of users into 7 groups (the number of clusters is obtained by the elbow method) class_rec.ipynb - recommendations for groups local_validation_groups.ipynb - combination of personal and group recommendations

Operating procedure:

1. Run data_split
2. Run user_class.ipynb, get a dataframe with the division of users into groups
3. Run class_rec.ipynb, get a dictionary with recommendations for groups
4. Run local_validation_groups.ipynb, get model prediction