SVC and LinearSVC have inconsistent decision_function shape #1490

Closed
amueller opened this Issue Dec 24, 2012 · 1 comment

Projects

None yet

2 participants

@amueller
scikit-learn member

In the binary case.
Not sure how that got across test_common.

@arjoly
scikit-learn member

Is it still the case?
It looks like a bug :-)

@amueller amueller added a commit to amueller/scikit-learn that referenced this issue Jul 18, 2014
@amueller amueller ENH Make all decision functions have the same shape. Fixes SVC and Gr…
…adientBoosting. Closes #1490.
71baf62
@amueller amueller added a commit to amueller/scikit-learn that referenced this issue Jul 18, 2014
@amueller amueller ENH Make all decision functions have the same shape. Fixes SVC and Gr…
…adientBoosting. Closes #1490.
af01125
@amueller amueller added a commit to amueller/scikit-learn that referenced this issue Jul 18, 2014
@amueller amueller ENH Make all decision functions have the same shape. Fixes SVC and Gr…
…adientBoosting. Closes #1490.
158fff0
@amueller amueller added a commit to amueller/scikit-learn that referenced this issue Jul 18, 2014
@amueller amueller ENH Make all decision functions have the same shape. Fixes SVC and Gr…
…adientBoosting. Closes #1490.
50b9f78
@amueller amueller added a commit to amueller/scikit-learn that referenced this issue Jul 19, 2014
@amueller amueller ENH Make all decision functions have the same shape. Fixes SVC and Gr…
…adientBoosting. Closes #1490.
e23d9c9
@amueller amueller closed this in #3434 Jul 19, 2014
@RolT RolT added a commit to RolT/scikit-learn that referenced this issue Sep 7, 2014
@amueller amueller split test_common.py into checks and test file.
move dataset generation into estimator_checks

Closes #2360. Fix tiebreaking.

some cleanups in common_test, speedup.

Make fit_transform and fit().transform() equivalent in nmf

slight speedup
some cleanup

Make everything accept lists as input.

Remove 'DeprecationWarning: using a non-integer number instead of an integer will result in an error in the future' warnings

Remove a couple more 'using a non-integer number instead of an integer DeprecationWarning'

Added a parameter check on n_estimators in BaseEnsemble to raise error if it is not strictly positive.

ENH Make all decision functions have the same shape. Fixes SVC and GradientBoosting. Closes #1490.

ENH better default for test for SelectKBest and random projection

MAINT tree compute feature_importance by default

pep8

ENH add label ranking average precision

DOC write narrative doc for label ranking average precision

DOC FIX error + wording

TST invariance testing + handle degenerate case

flake8

FIX use np.bincount

DOC friendlier narrative documentation

ENH simplify label ranking average precision (thanks @jnothman)

ENH be backward compatible for old version of scipy

Typo

pep8

DOC remove confusing mention to mean average precision

DOC improve documentatino thanks to @vene and remove mention of relevant labels

DOC more intuition about corner case

DOC add documentation to backported function

ENH less nested code

FIX encoding issue

DOC update what's new

MAINT deprecate fit_ovr, fit_ovo, fit_ecoc, predict_ovr, predict_ovo, predict_ecoc and predict_proba_ovr

ENH + DOC set a default scorer in the multiclass module

DOC typo + not forgetting single output case

scorer: add sample_weight support

COSMIT Use explicit if/else in scorer

TST default scorers with sample_weight

DOC Update What's New

MAINT split sklearn/metrics/metrics.py

DOC improve documentation and distinguish each module

ENH add a friendly warnings before deleting the file

TST: fix tests on numpy 1.9.b2

Refactor input validation.

remove check_arrays stuff and old input validation

ENH add allowed_sparse named argument for @ogrisel

FIX classes name in OvR

DOC mention doc-building dependency on Pillow

DOC link example gallery scripts rather than inline

COSMIT

DOC show referring examples on API reference pages

DOC ensure longer underline

DOC Fix example path

DOC fix 'Return' -> 'Returns'

FIX Py3k support for out-of-core example

DOC make neural networks example appear

Also make the absence of a README fail doc compilation

DOC add links to github sourcecode in API reference

DOC fix opaque background glitch when hovering example icons

FIX better RandomizedPCA sparse deprecation

added quantile strategy to dummy classifier

replaced quantile by percentile, added docstrings

switched back to quantile

final commit

forgot to save the file

removed blank lines, added default alpha param

added backport for axis keyword in scipy.scoreatpercentile

added quantile strategy to dummy classifier

replaced quantile by percentile, added docstrings

switched back to quantile

final commit

forgot to save the file

added backport for axis keyword in scipy.scoreatpercentile

removed scoreatpercentile axis dependency in the dummy tests

typo

fixed import

fixed axis keyword mistake

changed fix function name, fixed rebase bug

removed blank lines, added default alpha param

removed blank lines, added default alpha param

removed unnecessary import

docstring merge typo

updated doc (quantile dummy regressor)
1a376af
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment