Data Science Session: TabNet

This session shows how to:

• Train TabNet Classifier for a multi-task prediction
• Pre-train TabNet
• Visualize TabNet training metrics
• Interpret TabNet feature importances
• Perform data preparation for TabNet training (categorical feature encoding, filling missing values, etc.)
• Comparison of TabNet results with XGBoost results
• Check the similarity of feature distributions between train and test datasets

Pre-requirements

• python 3.8
• pip

Requirements

pip install -r requirements.txt

---

Speaker: Raid Arfua