This is a solution for DataFusion 2022 (VTB). This repo is a reworked version of Sberbank Pytorch-Lifestream (Apace 2.0 License, but repo was deleted). Practically all the files have been changed.
Embeddings for transactions are generated in main.py Embeddings for click streams are generated in main_cs.py Catboost model is trained in combine.py "container" directory contains inference script run.py
Solution for higher education prediction could be found in main_education.py
pip install pipenv
python -m pipenv install Pipfile