docs: update descriptions

[Planeshifter]

Description

What is the purpose of this pull request?

This pull request normalizes the NaN notation and summary-line punctuation in three @stdlib/stats/incr packages whose documentation had drifted from the convention used by the rest of the namespace.

Namespace summary

• Members analyzed: 163 (lib/node_modules/@stdlib/stats/incr/*); none autogenerated.
• Features analyzed: 15 — file tree, package.json top-level keys, package.json scripts keys, package.json stdlib keys, manifest.json shape, test/ naming, benchmark/ naming, examples/ naming, README heading set, README heading order, public signature, return kind, validation prologue, error construction, and JSDoc shape (param/returns/throws/example tags).
• Features with a clear majority (≥75%): 13. Twelve are at 100% conformance and yielded no outliers at all — notably, format is already used for 62/62 (100%) of members that construct an error, so no error-construction normalization was needed. README heading set was the only majority feature with outliers.
• Features without a clear majority (excluded): 2 — publicSignature (largest cohort 59/163 = 36.2%) and validationPrologue. Both vary legitimately: the namespace deliberately mixes zero-argument, window-size, mean-hint, and output-array accumulator shapes.

stats/incr is unusually well-kept. Of the seven candidate corrections that survived majority voting, four were dropped during validation, leaving three.

@stdlib/stats/incr/nancv

Normalizes the nancv README summary to match the package's own docstrings and the 95.8% convention across stats/incr. 'NaN' in ASCII single quotes was the only single-quoted occurrence of NaN in namespace prose; 69/72 nan* members of stats/incr (95.8%) backtick it. Also replaces "while ignoring" with the comma form ", ignoring", again matching 69/72 (95.8%), and bringing the README in line with nancv's own lib/main.js, lib/index.js, docs/repl.txt, and docs/types/index.d.ts, which already used the correct phrasing — the README was the package's sole deviating surface.

@stdlib/stats/incr/nanmmaxabs

Adds the missing comma before "ignoring" in the README description, matching the pattern used by 69 of the 72 nan* members of stats/incr (95.8%), which read "incrementally, ignoring NaN values." The package's own package.json description already carries the comma, as do the behaviorally identical siblings nanmmax, nanmmin, and nanmminabs. The JSDoc summary in lib/main.js carries a separate, larger deviation ("while ignoring NaN values") replicated across lib/index.js, docs/repl.txt, and docs/types/index.d.ts; that is left untouched here and warrants one coordinated pass.

@stdlib/stats/incr/nansumprod

Backticks NaN in the lib/main.js JSDoc summary to match the rest of the package. 69 of 72 nan* accumulators in @stdlib/stats/incr (95.8%) already backtick NaN in their summary line, and every other documentation surface in this package — README.md, lib/index.js, docs/repl.txt, docs/types/index.d.ts — already does so; lib/main.js was the sole holdout. Strictly reduces intra-package inconsistency; introduces none.

Related Issues

Does this pull request have any related issues?

None.

Questions

Any questions for reviewers of this pull request?

Yes — one, on the drift that was deliberately not fixed here.

nanmmaxabs, nanmstdev, and nanmvmr each replicate the same deviating summary sentence across five surfaces: README.md, lib/main.js, lib/index.js, docs/repl.txt, and docs/types/index.d.ts. In those three packages lib/main.js currently agrees with docs/repl.txt and docs/types/index.d.ts, so editing lib/main.js alone would trade an existing README-vs-JSDoc mismatch for a new JSDoc-vs-declaration mismatch. nanmstdev is the worst case: the namespace TOC entry at lib/node_modules/@stdlib/stats/incr/README.md:121 is generated from its README summary line by _tools/markdown/namespace-toc, and lib/node_modules/@stdlib/stats/incr/docs/types/index.d.ts:2419 mirrors the unbackticked form, so a correct fix also requires regenerating the TOC (make markdown-namespace-tocs).

Should that five-surface pass be folded into this pull request, or handled separately?

Other

Any other information relevant to this pull request?

Validation

Every correction was checked as follows before being applied:

• Structural extraction — file tree, package.json key sets, README heading sequences, and test/benchmark/examples file naming, over all 163 members.
• Semantic extraction — public signature, return kind, validation prologue, error construction, JSDoc tag shape, and require()-derived dependencies, over all 163 members.
• Three-agent drift validation — an independent semantic review (does the deviation reflect a genuine difference in what the package computes?), a cross-reference review (do tests, benchmarks, examples, type tests, or repo tooling depend on the deviating text?), and a structural review (is the majority pattern the thing that should actually be applied?).
• No test, benchmark, example, type test, or _tools fixture references any of the changed text. remark-lint-expected-html-sections treats the notes section as optional, and stdlib/jsdoc-maximum-line-length is configured off, so neither changed line carries lint risk.
• nansumprod/lib/main.js was re-parsed with node --check and exercised at runtime after editing (2*3 + 1*5 = 11 with the NaN pair correctly skipped).

Deliberately excluded

• Intentional deviations. stats/incr/count lacks the ## Notes section present in 158/163 members (96.9%), but incrcount's accumulator never reads its argument — it increments on arguments.length alone — so the majority bullet's warning that a NaN input poisons all future values would be factually false there. Two independent reviews reached this conclusion separately.
• Outliers needing a coordinated pass. The lib/main.js JSDoc corrections for nanmmaxabs, nanmvmr, and nanmstdev, and the nanmstdev README correction — see Questions above.
• Outliers whose fix is not mechanical. nancount and nancovariance are also missing ## Notes. Content exists to lift, but only 51 of 160 ## Notes sections in the namespace consist of the standard bullet alone; the other 109 carry package-specific prose. nancovariance's README additionally deviates in ways outside this analysis (#### incrnancovariance with no signature, #### accumulator(`[x, y]`) instead of #### accumulator( \[x, y] ), unbackticked NaN in the Usage prose), which wants one scoped README pass rather than a drive-by section add.
• Features without a clear majority. publicSignature and validationPrologue, per the namespace summary above.
• Generator-owned surfaces. docs/repl.txt, docs/types/index.d.ts, README ## See Also blocks, and the commented-out equation <div> blocks were treated as out of scope.
• Already in flight. Two findings were dropped as duplicates of open work: the ## Notes gap in covmat and pcorrmat (both carry an empty <section class="notes"> with no heading) is proposed by docs: add NaN notes to stats/incr/covmat and stats/incr/pcorrmat #12693, and the backticked-NaN package.json descriptions in nanmmda and nanmmpe are corrected by chore: address review findings in new stats/incr/nan* packages #13602.
• Not a fix we would make. nanmmaxabs is the only one of the 19 nanm*( W ) delegating accumulators that re-validates W itself; the other 18 (94.7%) rely on the delegate to throw. Conforming would mean removing a validation check, which is behavior-adjacent, so it was left alone.

A note on @throws without throw

44 nan* packages document @throws while containing no throw statement. This is not drift and was not flagged: each delegates construction to its non-nan sibling (e.g. incrnanmmean( W ) calls incrmmean( W )), so the documented error is genuinely thrown and propagates.

Checklist

Please ensure the following tasks are completed before submitting this pull request.

• Read, understood, and followed the contributing guidelines.

AI Assistance

When authoring the changes proposed in this PR, did you use any kind of 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 pull request was authored by Claude Code running an automated cross-package drift-detection routine. The routine extracted structural and semantic features from all 163 stats/incr packages, computed the majority pattern per feature at a 75% threshold, and flagged deviating packages. Candidate corrections were then reviewed by three independent agents; four of seven candidates were rejected at that stage and are documented under "Deliberately excluded" above. The three changes that remain are one-line documentation edits. Reviewers should scrutinize the conformance percentages and the judgement that each deviation is unintentional rather than meaningful; the underlying measurements are reproducible from the namespace itself.

---

🤖 Generated with Claude Code

https://claude.ai/code/session_01FFCQ8XaJ5NUDv9oDpHF2Vj

---

Generated by Claude Code

[stdlib-bot]

Coverage Report

Package	Statements	Branches	Functions	Lines
stats/incr/nancv	$\\color{green}141/141$ $\\color{green}+100.00\\%$	$\\color{green}11/11$ $\\color{green}+100.00\\%$	$\\color{green}2/2$ $\\color{green}+100.00\\%$	$\\color{green}141/141$ $\\color{green}+100.00\\%$
stats/incr/nanmmaxabs	$\\color{green}148/148$ $\\color{green}+100.00\\%$	$\\color{green}9/9$ $\\color{green}+100.00\\%$	$\\color{green}2/2$ $\\color{green}+100.00\\%$	$\\color{green}148/148$ $\\color{green}+100.00\\%$
stats/incr/nansumprod	$\\color{green}135/135$ $\\color{green}+100.00\\%$	$\\color{green}8/8$ $\\color{green}+100.00\\%$	$\\color{green}2/2$ $\\color{green}+100.00\\%$	$\\color{green}135/135$ $\\color{green}+100.00\\%$

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