op-mat-correlation - Pulse Documentation

name: op-mat-correlation
description: Correlation matrix of a vector's numeric members — Pearson (default), Spearman or Kendall tau-b via params.method — listwise or pairwise, weighted; one MatrixResult per spec (per group bucket when grouped), no p-values.
kind: operator
category: MAT
operator: MAT_CORRELATION
type: reference
applies_to: process, compose, predict
examples_tags: [matrix, correlation-analysis]

Slot: matrices[i] {type, vector | fields, params, weight, encoding}; members as for op-mat-covariance.

Use when

How closely every pair of numeric fields moves together, in a line or in rank order, as one square table of values from -1 to 1.

Questions it answers:

Which of these ten rating items go together most strongly?
How do the sensor readings line up with one another, pair by pair?

Use something else:

TEST_PEARSON_R when you need a p-value or confidence interval for one pair.

Params

Name	Type	Default	Description
`method`	`pearson` \| `spearman` \| `kendall`	`pearson`	Rank methods: buffered, serial.
`missing`	`listwise` \| `pairwise`	`listwise`	Pairwise: per-pair rows (rank methods re-rank per pair).
`max_drop_share`	0–1	none	Listwise drop-share warning.
`summary.top_pairs`	int ≥ 1	none	k strongest pairs.
`weight`	slot weight	inherited	Pearson: both kinds; rank: frequency only. `null` opts out.
`encoding`	`full` \| `upper`	`full`	`upper`: row r holds p − r cells.

Inputs

Integer / float members; packed_bool under coerce: "binary".

primary: r (ρ / τ-b under a rank method) in [−1, 1], diagonal 1; a zero-spread member’s row and column are null. scalars.determinant (null unless PD); pairwise auxiliary.n (pair N); warnings. vectors.top_pairs [{row, col, r, n}]: by |r| desc, ties in axis order, no diagonal / null pairs.

Reading the output

primary.values: Each off-diagonal cell is the correlation of its row and column members under params.method, from -1 to +1. pearson (default): r, how closely the two follow a straight line together; spearman: rho, r on the members’ ranks, how steadily one rises or falls with the other; kendall: tau-b, the share of agreeing minus disagreeing row pairs, tie-adjusted. 0 means no such link. The diagonal is 1.
- Bands (Cohen (1988), absolute value): very small below 0.1; small 0.1 to 0.3; medium 0.3 to 0.5; large 0.5 and above.

Components

components.matrices[i]: n, n_null, n_listwise_dropped; pairwise min_pair_n / max_pair_n; weighted sum_weights, n_eff, n_weight_invalid.

Gotchas

Each cell is the matching two-field test (Pearson, Spearman, Kendall) over its pair’s rows. No p-values (auxiliary.p arrives with U28): run the test per pair. Weight-0 rows count in n, add no mass.
Rank methods: predict streamable / mergeable false; a probability weight is PULSE_WEIGHT_UNSUPPORTED. Kendall τ-b runs smaller than ρ.
Pairwise r at p ≥ 3 can be non-PSD (predict pairwise_psd_risk) → PULSE_MATRIX_NOT_PSD; also _INSUFFICIENT_N, _ZERO_VARIANCE, _LISTWISE_HEAVY_DROP.
Refusals and grouping: matrix-results.

See

pulse_examples_search tags=[matrix, correlation-analysis]
Skills: matrix-results, op-mat-covariance, weighting

Keyboard shortcuts

Pulse Documentation