docs: update parameter type and add keyword

[Planeshifter]

Description

This pull request aligns @stdlib/stats/strided/distances/dcorrelation with the seven sibling packages in the namespace.

Namespace summary

• Target: @stdlib/stats/strided/distances
• Members analyzed: 8 (dchebyshev, dcityblock, dcorrelation, dcosine-distance, dcosine-similarity, deuclidean, dminkowski, dsquared-euclidean)
• Features extracted: complete file tree, package.json shape (top-level keys, scripts, directories, keywords), manifest.json shape, README heading sequence, test/ benchmark/ examples/ file-name patterns, semantic features (public signature, JSDoc shape, dependency set, error construction, validation prologue, return kind) from lib/main.js, the per-package wrapper, and lib/ndarray.js.
• Features with clear majority (≥75% = 6/8): identical 33-file structure (100%); package.json top-level key set (100%); directories keys (100%); empty scripts (100%); manifest.json keys confs/fields/options (100%); 14 universal keywords at 100% (array, distance, double, float64, float64array, math, mathematics, statistics, stats, stdlib, stdmath, strided, strided array, typed); README H2 set (Usage, Notes, Examples, C APIs, 100%); H3 set (Usage, Examples, 100%); metric keyword at 7/8 = 87.5%; @param {NonNegativeInteger} N JSDoc in the lib wrappers and ndarray.js at 7/8 = 87.5%; full strideX, offsetX, strideY, offsetY parameter names in the README C-API ndarray H4 heading at 7/8 = 87.5% (corrected then reverted — see Dropped fixes below).
• Features without clear majority (excluded from drift detection): REPL phrasing Stride length of/for; TypeScript declaration phrasing (the canonical TS form `x` stride length differs by word order from the JS stride length of x`` form, so a simple of → for substitution would not produce the sibling pattern); description verb `Compute the …` at 100% (unchanged — PR docs: updae descriptions in stats/strided/*abs package descriptions #12310's `Compute → Calculate` standardization across the parent `stats/strided` namespace explicitly excluded the `distances` sub-namespace, and within this sub-namespace the verb is consistent); metric-specific keywords (e.g. `cosine`, `euclidean` — package-specific, no majority).

@stdlib/stats/strided/distances/dcorrelation

dcorrelation lagged the other seven stats/strided/distances packages on its JSDoc and package metadata. Its @param {PositiveInteger} N JSDoc across lib/dcorrelation.js, lib/dcorrelation.native.js, lib/ndarray.js, and lib/ndarray.native.js contradicts the tested N <= 0 → NaN contract — 7/8 siblings annotate NonNegativeInteger, and tests at test/test.dcorrelation.js:64-78 and test/test.ndarray.js:64-78 already exercise the N=0 and N<0 paths. The package.json keywords array was missing metric, the only universally-present semantic tag across the seven sibling distance packages (dchebyshev, dcityblock, dcosine-distance, dcosine-similarity, deuclidean, dminkowski, dsquared-euclidean all carry it).

Validation

Three independent validation agents reviewed each candidate fix before any edit was applied:

1. Semantic-review (Opus): read dcorrelation/lib/dcorrelation.js, lib/ndarray.js, both .native.js counterparts, both test suites, and the same files for deuclidean and dcosine-distance for comparison. Verdict: confirmed-drift on the JSDoc N-type retype; the function explicitly handles N <= 0 by returning NaN, exactly matching siblings, so the PositiveInteger annotation has no semantic basis.
2. Cross-reference (Opus): searched for downstream consumers of the literal pre-fix strings across the wider lib/node_modules/@stdlib/ tree. Verdict: safe-to-fix on the keyword addition (no tooling treats "metric" as a semantically meaningful discovery index), the README heading correction (no test, benchmark, or example references sx, ox, sy, oy), and the JSDoc retype (no runtime check, test, or .d.ts consumes the JSDoc type tag). Verdict: needs-human on a broader REPL/TS wording cleanup — the candidate conflated lib/ndarray.js's universal stride length of form (which is not drifted) with docs/repl.txt's drifted Stride length of form and with docs/types/index.d.ts's entirely different `x` stride length word order; that bucket was dropped.
3. Structural-review (Sonnet): confirmed that "metric" semantically applies to a correlation-distance package — the correlation distance d_c = 1 - r(X, Y) is a valid pseudometric (non-negative, symmetric, triangle inequality), and the sibling pattern applies the keyword uniformly across the group regardless of strict-metric vs pseudometric distinction (dcosine-similarity, which is not itself a metric, also carries the keyword).

Dropped fixes (after agent validation passed)

• README ## C APIs ndarray H4 heading: initially replaced abbreviated sx, ox, sy, oy placeholders with the full strideX, offsetX, strideY, offsetY form (matching all 7 siblings) in commit 4415b3397. CI's Run changed examples job flagged the resulting 91-character heading against remark-lint's maximum-heading-length: 80 rule. Every sibling heading in the namespace exceeds the same limit (89-97 characters), but the rule only runs on PR-touched READMEs; the violation has accumulated silently. Reverted in commit a000843e6. The README's argument list (lines 241-245) and the literal C signature in the same file (line 248) already use the full parameter names, so readers see the correct names; only the heading retains the abbreviation. The namespace-wide heading-length issue is a separate concern.

Other deliberately excluded

• REPL phrasing cleanup (Stride length of → Stride length for in docs/repl.txt) and the TypeScript declaration rewording (stride length of x→ x stride length , `starting index of `x → starting index for x`` in docs/types/index.d.ts) — agent Feature/sum-series #2 flagged the candidate as ambiguous; the underlying drift is real but the fix requires file-by-file disambiguation rather than a single substitution. Logged for a future targeted pass.
• Removal of the self-referential "dcorrelation" keyword (1/8 occurrence; siblings exclude the d-prefixed package name) — outside the validated scope and would need a separate chore:-level run.
• dcorrelation/src/main.c @param C-comment phrasing — real drift, but outside the agent-validated scope of this run.

Related Issues

This pull request has no related issues.

Questions

No.

Other

• Final diff: 5 lines across 5 files (one package.json keyword insertion, four JSDoc @param N retypes).
• No tests, examples, benchmarks, REPL fixtures, TypeScript declarations, or REPL docs were modified.
• Full local report (member list, feature distributions, per-agent verdicts, dropped findings, future-run notes) saved at ~/drift-reports/drift-stats-strided-distances-2026-06-09.md outside the repo working tree.

Checklist

• Read, understood, and followed the contributing guidelines.

AI Assistance

• Yes
• No

If you answered "yes" above, how did you use AI assistance?

• Code generation (e.g., when writing an implementation or fixing a bug)
• Test/benchmark generation
• Documentation (including examples)
• Research and understanding

Disclosure

This PR was authored by Claude Code running the cross-package API drift detection routine on the randomly-selected @stdlib/stats/strided/distances namespace. Structural features (file tree, package.json shape, README headings, etc.) were extracted across all eight members; semantic features (JSDoc shape, public signature, dependency set, validation prologue, return kind) were read directly from each member's lib/main.js, per-package wrapper, and lib/ndarray.js. Three independent validation agents (opus semantic-review, opus cross-reference, sonnet structural-review) reviewed each candidate fix before any file was edited. A human should audit before promoting from draft.

---

@stdlib-js/reviewers

[stdlib-bot]

Coverage Report

Package	Statements	Branches	Functions	Lines
stats/strided/distances/dcorrelation	$\color{green}366/366$ $\color{green}+100.00\%$	$\color{green}15/15$ $\color{green}+100.00\%$	$\color{green}4/4$ $\color{green}+100.00\%$	$\color{green}366/366$ $\color{green}+100.00\%$

The above coverage report was generated for the changes in this PR.