ml.gradient_boosting

Histogram gradient boosting regression or classification (scikit-learn). Writes a SpadeModel with an ONNX graph that models.predict scores without Python; before the model is written, the graph scores the training rows and must agree with the fitted model, or the block fails. Early stopping is off, so the number of trees never depends on a random validation split.

Kind: standard
Network: no

Inputs🔗

Name	Type	Description
`data`	file (Parquet or CSV)	Training table: one row per observation, with the target, the predictor columns, and any weights or id columns.
`target`	string	The response column.
`features`	string	Column selector for the predictors: names, globs and exclusions, e.g. `B,MB,-B0`.
`task`	string	`regression` (default) predicts a number; `classification` predicts the target's classes, with a probability for each.
`id_columns`	string	Column selector for identifier columns, never used as predictors; kept as text when reading CSV. Default empty.
`weights`	string	Column of non-negative observation weights. Empty fits unweighted.
`categorical`	string	Column selector for numeric-coded predictors to treat as categorical. Categoricals are one-hot encoded, one column per level. Default empty.
`na_action`	string	`fail` (default, naming the columns) or `drop`: what to do with a row that has a missing value in a model column.
`seed`	number	Seed for the model's randomness and for fold assignment; the same seed gives the same model. Default `0`.
`cv_folds`	number	K-fold cross-validation (`0` for none). The model is always fitted on all rows. Default `0`.
`cv_group`	string	Column whose values are kept together in one fold (e.g. a county id), so spatially related rows do not leak between folds. Default empty.
`learning_rate`	number	Shrinkage applied to each tree. Default `0.1`.
`max_iter`	number	Boosting iterations (trees per class). Default `100`.
`max_depth`	number	Maximum tree depth. Leave empty (the default) for no limit beyond `max_leaf_nodes`.
`max_leaf_nodes`	number	Maximum leaves per tree. Default `31`.
`min_samples_leaf`	number	Minimum rows in a leaf. Default `20`.
`l2_regularization`	number	L2 penalty on leaf values. Default `0`.

Outputs🔗

Name	Type	Description
`model`	file (SpadeModel)	The fitted model with its ONNX graph, for `models.predict`.
`fit`	json	Rows used and dropped, the arguments that shaped the fit, and `metrics`: `rmse`, `mae`, `r_squared` (regression) or `accuracy`, `log_loss` (classification), with `cv_` versions when cross-validated.

Example🔗

- id: "@fit"
  name: ml.gradient_boosting
  inputs:
    - "@plots"
  args:
    target: "present"
    features: "elevation_m,canopy_cover_pct,forest_type"
    task: "classification"
    cv_folds: 5