Options

Options are a single JSON object in the last argument. Unknown keys and wrong-typed values are rejected with PREDICT_ERR_OPTIONS, so a typo fails loud rather than being ignored.

forecast

Key	Type	Default	Meaning
model	string	auto	A model id to pin, or auto (the default: best of the stats plus eligible registered students, per series).
confidence_level	number (0,1)	0.9	Nominal coverage of the interval.
interval_method	residual | conformal	residual	Interval construction.
folds	integer	20	Rolling origins used by auto / conformal.
gap	integer	0	Leakage guard between train and target.
candidates	string[]	stat models + eligible registered students	Narrows the auto pool.
context_limit	integer	4096	Cap on points fed to the model (the most recent are kept; a capped series reports status truncated).

Aggregate rules: the options argument and horizon must be the same value on every row of a group, and a SELECT string passed as the first argument is rejected with an error explaining that forecast is an aggregate over your rows. There are no column-naming keys: the argument positions carry the columns and GROUP BY carries the series split.

detect_anomalies

model and context_limit as above, except that model defaults to theta-classic (the residual detector) and context_limit defaults to 0: the detector scores the whole series unless you cap it. Plus:

Key	Type	Default	Meaning
anomaly_prob_threshold	number (0,1)	0.99	Probability above which a point is flagged.

The same aggregate rules apply.

backtest

An aggregate over your rows, like forecast. confidence_level, interval_method, folds, gap, and context_limit as for forecast. model differs: it must be a statistical model id or auto, and it defaults to theta-classic, not auto; a distilled student competes inside auto but cannot be pinned here. There are no column-naming keys: the argument positions carry the columns and GROUP BY carries the series split.

fit

Trains a native tabular student over your rows; the label is the last positional argument, so there is no target option.

Key	Type	Default	Meaning
kind	gbt | tree	gbt	Student architecture.
task	classify | regress	inferred	Inferred from the label: integer or text is classify, real is regress.
register	string	none	Register the model under this id and return the id; without it, fit returns a model blob.

predict

A scalar: predict(model, f1, ..., fN [, options]). model is a registered id or a fit() blob; features are positional.

Key	Type	Default	Meaning
proba	0 | 1	0	Return a {"prediction": "1", "confidence": 0.98} JSON document instead of the bare label.

predict_batch

The batched and in-context serving path, predict_batch(train_query, apply_query [, options]).

Key	Type	Default	Meaning
target	string	required for in-context models	Column in train_query to learn. A served model (train_query = NULL) needs no target.
task	classify | regress	inferred	Prediction task.
model	string	knn5-incontext	An in-context model, a registered native student id (trained by distill_predict or fit), or a registered onnx id. To serve an already-trained model (a native student or onnx), pass train_query = NULL; only in-context models take a train_query.
device	cpu | gpu	cpu	Inference device (onnx build).
precision	string	model default	Inference precision (onnx build).
accept_license	0 | 1	0	Accept a license-tagged model.

distill_predict

target and student_id are required. Feature columns must be numeric: encode categorical text before distilling (the in-context knn5-incontext handles text features itself; the distiller does not).

Key	Type	Meaning
target	string	Target column. Required.
task	classify | regress	Task. Inferred when omitted.
student_id	string	Id to register the student under. Required.
student_kind	tree | gbt | mlp	Student architecture. Default tree (soft-label distillation implies gbt).
teacher	string	A registered model that relabels the training rows first.
proba, classes	string[]	Soft-label distillation: per-class probability columns and their class labels.

distill_forecast

context, horizon, and student_id are required. Without a teacher, each train_query row is one training window laid out by position: context input columns, then the horizon continuation columns (or horizon * nquant quantile columns when quantiles is passed); the column count must match exactly.

Key	Type	Meaning
student_id	string	Id to register the student under. Required.
teacher	string	Teacher forecast model (onnx build): train_query then yields (series_key, value) rows and the teacher labels sliding windows.
context	integer	Input window length. Required; max 4096.
horizon	integer	Forecast length the student is trained for. Required.
quantiles	number[]	Quantile levels to distill from a teacher’s quantile fan. Default: a point (median-only) student.
hidden	integer	Residual-net hidden width, 0 to 2048 (0 = pure linear student). Default 256.
epochs, lr		Training epochs (default 1500) and learning rate (default 0.005).