Feature engineering

[kurtosis111]

1. Correlation of features from the same tag is high: https://www.kaggle.com/code/lifuhaha/analysis-of-weight-by-symbol-id
2. Feature clustering: https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting/discussion/542209
3. Baseline notebooks: https://www.kaggle.com/code/motono0223/js24-inference-gbdt-with-lags-singlemodel

[kurtosis111]

[kurtosis111]

correlation of lag with responder_3:

[kurtosis111]

This conclusion is only based on the train_parquet 0, 1, 2.