Model Connectors
This section explains which models are available in the Myst Platform
Model | Provider | Documentation |
---|---|---|
LightGBM | LightGBM | Documentation |
MLP Regression | PyTorch | Documentation and tutorial |
Elastic Net | Scikit-learn | Documentation |
Extra Trees Regression | Scikit-learn | Documentation |
Linear Regression | Scikit-learn | Documentation |
Logistic Regression | Scikit-learn | Documentation |
Random Forest Regression | Scikit-learn | Documentation |
XGBoost | XGBoost | Documentation |
NGBoost | NGBoost | Documentation |
ARIMAX | statsmodels | Documentation |
Note on Missing Data
For all models, when training, observations are dropped when targets are null. For MLP, Elastic Net, Extra Trees Regression, Linear Regression, and Random Forest Regression, observations with null features are also dropped.
XGBoost and LightGBM offer parameters to control whether they fit or predict on null features.
In general, nulls at prediction time result in null values at that timestamp in the forecast.
Updated about 2 years ago