otus-ht5

Task: Churn analyse

Implemented:

• Design a criteria for churned users and create labels;
• Design new features;
• Data aggregation;
• Train three classification models and choose the best one;
• Select best features;
• Result interpretation.

Files:

• explore.ipynb - Data analyse and transformations.
• models.ipynb - Models training and interpretations.

The dataset for the project can be found here: https://www.kaggle.com/sharthz23/sna-hackathon-2019-collaboration?select=train.

The project is written on Scala, please use Almond to run it: https://almond.sh/