Skip to content

CHANGELOG: label the v0.3.0 section and point the links at this repository - #13

Merged
luizfernandolj merged 1 commit into
masterfrom
changelog-v030-label
Nov 9, 2026
Merged

luizfernandolj merged 1 commit into
masterfrom
changelog-v030-label

Conversation

@luizfernandolj

Copy link
Copy Markdown
Collaborator

Three small defects in CHANGELOG.md, found while writing the Unreleased entries in #12.

The v0.3.0 section was still labelled Unreleased

  7:## [Unreleased]
 95:## [v0.5.0]
236:## [v0.4.0]
390:## [Unreleased] — v0.3.0

It was released, so the file had two headings matching ## [Unreleased], and anything looking for the current one can find the wrong block. That is not hypothetical: a script of mine checking the Unreleased section reported 33 entries and sections that do not exist there, because its range restarted at line 390 and swept half the file.

Its position was already right — below v0.4.0 in the descending list. Only the label was wrong, so this renames it to ## [v0.3.0].

That makes it a markdown reference link, and there was no definition for it, so one is added against v0.2.1, the tag immediately before v0.3.0. Every ## [...] heading in the file now resolves:

cabecalhos: ['Unreleased', 'v0.5.0', 'v0.4.0', 'v0.3.0']
sem definicao: nenhum

The compare links pointed at the old repository

They still said github.com/luizfernandolj/mlquantify. GitHub redirects transferred repositories, so they resolved — this is staleness, not breakage — but the project lives at coenlab/mlquantify now.

The Unreleased link compared from the wrong tag

[Unreleased] compared from v0.5.0. Everything in that section landed after v0.5.1, so the range also pulled in v0.5.1's own changes. It now compares from v0.5.1.

Not included

v0.4.1 and v0.5.1 have no sections in this file at all — only Unreleased, v0.5.0, v0.4.0 and v0.3.0 exist. Writing those retroactively means reconstructing release notes from history, which is a different job from fixing a label.

Markdown only; no code and no behaviour change.

🤖 Generated with Claude Code

…itory

The v0.3.0 section was headed "[Unreleased] - v0.3.0". It was released, so
the file had two headings matching "## [Unreleased]" and anything looking
for the current one could find the wrong block. Its position was already
correct, below v0.4.0 in the descending list; only the label was wrong.

Renaming it to "## [v0.3.0]" makes it a reference link, which needed a
definition, so one is added against v0.2.1, the tag before it.

The compare links still pointed at github.com/luizfernandolj/mlquantify.
GitHub redirects transferred repositories so they resolved, but the project
lives at coenlab/mlquantify now.

The Unreleased link compared from v0.5.0. Everything in that section landed
after v0.5.1, so the range pulled in v0.5.1's changes as well; it now
compares from v0.5.1.

Markdown only.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
@luizfernandolj
luizfernandolj merged commit 8546fec into master Nov 9, 2026
14 checks passed
luizfernandolj added a commit that referenced this pull request Nov 11, 2026
#14 removes the estimator from MLPE: MLPE(LogisticRegression()) now raises
TypeError, cv, stratified and shuffle are gone from the signature, and there
is no estimator_ after fit. Anyone upgrading without reading the diff finds
that at runtime, and the Unreleased section had no Breaking Changes heading
at all -- although the v0.5.0 entry already uses that section for exactly
this kind of change, the evaluation-metrics argument reorder.

The entry leads with what breaks and how to migrate (drop the argument --
MLPE()), and says plainly that the estimate does not change: it was always
the training prevalence read from the labels. The breakage is in the
signature, not in the numbers.

Then the measurement that justified it. CC, which trains one classifier and
nothing else, cost 0.0523 s against MLPE's 0.0571 s over 142 datasets, and
QuaPy's MLPE, which is non-aggregative and counts labels in two lines, cost
0.0001 s -- a ratio of 347x for work the estimate never used. fit now takes
0.55 ms instead of about 57 ms. Against QuaPy the training-time ratio falls
from 347x to 4.99x, training memory from 6.90x to 2.04x, estimation memory
from 1.90x to 1.00x, and training plus estimation from 8.51x to 1.38x. The
whole method over 142 datasets takes 50 s of CPU against 40 min.

Every figure was checked against q2_custo.csv of the two runs before being
written.

#13 is not logged: it changed this file's own section label and compare
links, which is not a user-visible change to the library.

Markdown only.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant