Hi, we are the team of AndyIsMyBoss. In this repo, we share the code of how we achieved 2nd prize in 2021 Winter Esun AI Competition - Credit Card Consumption Tag Recommendation. To learn more information about this competition, please refer to Competition Website. More details about the method we used in this code, please refer to Article, Video, and Presentation Slide.

• python 3.8.5
• numpy
• pandas
• csv
• matplotlib
• tqdm
• pickle
• scipy
• torch
• sklearn

• RNN Model.ipynb
  • The method we used to achieve 2nd prize in this competition.
  • In this file, it includes the process of feature engineering, model construction, and training method.
• Ensemble.ipynb
  • The method of soft-ensemble to achieve higher score on leaderboard.
• Ranking Loss.py
  • The PyTorch version of many different ranking loss.
  • For Tensorflow version, please refer to TF Ranking.
• Transformer.py
  • The transformer model we implemented.
• DASALC.py
  • We implemented the method published in 2021 ICLR by Google.
  • For more details, please refer to Are neural rankers still outperformed by gradient boosted decision trees?

Chieh-Fang Teng:
     + [email protected]
     + [email protected]