hannawong / WSDM22_IQIYI

For WSDM 2022 IQIYI User Retention Score Prediction

WSDM 2022 IQIYI User Retention Score Prediction Solution

Directory

Organize the data as the following directory tree:

- data
- raw_data
    - app_launch_logs.csv
    - user_interaction_data.csv
    - user_playback_data.csv
    - user_portrait_data.csv
    - video_related_data.csv
    - test-a.csv
- feature.py
- model_tools.py
- model.py
- train.py

How to run this code

Firstly run the following command for data preprocessing

python feature.py

Then run the following command for training & evaluating & generating submission file

python train.py

Ways to improve 💡

• Use user_portrait_data.csv as user-side features. Make sure to encode each categorical data in one-hot manner and discretize each numerical data by equal-depth binning, and then embed all the features with embedding table.

• Combine user_interaction_data and user_playback_data with video_related_data to formulate user sequential behavior, and use model like DIEN+auxillary loss, Transformer, MIMN, or other state-of-the-art methods to model such sequential behavior.