op-attr-reg-fitted - Pulse Documentation

name: op-attr-reg-fitted
description: Per-row fitted value ŷᵢ = Xᵢ · β + β₀ from an OLS refit during the attribute prepass.
kind: operator
category: ATTR
operator: ATTR_REG_FITTED
type: reference
applies_to: process, compose, predict
examples_tags: [regression, ols, buffered-pipeline]

Attributes emit row-level scalars; they do not produce Response.Components.

Use when

Adds to every row the value a straight-line model predicts for its target from its predictors.

Questions it answers:

What order value would we expect for each customer given their visits and tenure?
Which stores sell above or below what their size and footfall predict?

Use something else:

REG_OLS when you want the coefficients, R-squared and p-values of the model.
REG_GLM when the target is yes/no or a count.

Params

Name	Type	Description
`Target`	string	Dependent field (required).
`Predictors`	[]string	Independent fields (required).
`Penalty`	enum	`""`, `l1`, `l2`, `elasticnet`.
`Alpha`	float	Regularization strength.
`L1Ratio`	float	Elasticnet mix.
`weight`	slot weight	`null` opts out; REG_OLS’s kinds: WLS refit; residual raw y − ŷ.

Inputs

Param	Accepted field types
`Target` / `Predictors`	numeric (no `decimal128`)
`Label`	required — new column name

Output

One float64 per record — the model’s prediction ŷᵢ. Null (never 0) when any predictor is null, so a downstream aggregation counts the row in n_null; a null target alone still gets ŷᵢ.

Reading the output

value: The target value the fitted straight-line model predicts for the row from its predictors: intercept + sum(coefficient * predictor), in the target’s units. The model is fitted by least squares to the rows that passed the filters and have the target and every predictor.
- Caveat: The model describes association in these rows: a fitted value is what rows with these predictor values look like here, not what would happen to a row if a predictor were changed; a causal reading needs a design that supports it.

Gotchas

Two-pass: prepass refits independently per ATTR_REG_* slot (Option A — no fit sharing).
Mergeable per-shard; predict reports streamability.
Penalized fits (l1/l2/elasticnet) reuse the same machinery; mis-tuned Alpha/L1Ratio shrinks coefficients toward zero.

See

pulse_examples_search tags=[regression]
Skills: regression-modeling, attribute-composition, op-attr-reg-residual

Keyboard shortcuts