Pulse Documentation

name: op-attr-scale-score
description: Per-row rating-battery score — the valid-item mean or prorated sum across a vector of members, with reverse-keyed items and a min_valid floor.
kind: operator
category: ATTR
operator: ATTR_SCALE_SCORE
type: reference
applies_to: process, compose, predict
examples_tags: [feature-engineering, multivariate, streaming-friendly]

Use when

Adds to every row its score on a rating battery: the average of its answered items or a prorated total, reverse-worded items flipped.

Questions it answers:

• What is each respondent’s satisfaction score across the five satisfaction items, counting only people who answered at least four?
• What is each patient’s symptom total when two of the items are worded the other way round?

Params

• stat — mean (default) or sum: ALWAYS prorated (valid mean × item count); no raw sum.
• min_valid — integer 1…p, default p: fewer valid items → null, never 0.
• reverse — member LABELS, scored lo + hi − x; needs a range.
• scale_min / scale_max — both or neither; else a kind: scale vector’s bounds. No range + reverse → SERVICE_VALIDATION.

Slot: exactly one of vector (a vectors entry) / fields; no field. Default label ATTR_SCALE_SCORE_<vector>.

Inputs

Members: u4…u64, f32/f64; source columns only.

Output

One f64 per record on the item scale (sum: × p).

Components

components.attributes entry: floor {n, n_null} (present / null scores) + n_out_of_range (values), n_rows_out_of_range (rows) only with a range. Mergeable; no weighted keys.

Gotchas

• With a range, a value outside it is MISSING (never clamped): one PULSE_ATTR_SCALE_OUT_OF_RANGE warning, same counts. MAT_RELIABILITY refuses that data instead.
• Row-local: streams, fuses; weight ignored.

See

• pulse_examples_search tags=[feature-engineering, multivariate]
• Skills: attribute-composition, multivariate-design