Built an ensemble model on the Spotify dataset to determine the popularity of songs and study feature importance using SHAP.
The ensemble model includes:
- "ridge"
- "random forest"
- "XGBoost"
- "LightGBM"
- "CatBoost"
The predicted popularity is 50.07, which is close to true popularity 50.
The predicted popularity is 38.71, which is close to true popularity 41.