From ad7c628d974063e31eb749e83816ef15f54f7ae4 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Matthias=20K=C3=BCmmmerer?= Date: Fri, 24 Nov 2023 14:45:13 +0100 Subject: [PATCH] Remove buggy and uncessary crossval class MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The functionally is very simple and should, IMO, not be automatic. Signed-off-by: Matthias Kümmmerer --- CHANGELOG.md | 2 +- pysaliency/baseline_utils.py | 10 ---------- 2 files changed, 1 insertion(+), 11 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index 2a427f7..c29ce6b 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -3,7 +3,7 @@ * 0.2.22 (dev): * Bugfix: The NUSEF dataset scaled some fixations not correctly to image coordinates. Also, we now account for some typos in the dataset source data. - * Feature: CrossvalMultipleRegularizations, CrossvalGoldMultipleRegularizations and GeneralMixtureKernelDensityEstimator in baseline utils (names might change!) + * Feature: CrossvalMultipleRegularizations and GeneralMixtureKernelDensityEstimator in baseline utils (names might change!) * Feature: DVAAwareScanpathModel * Feature: ShuffledBaselineModel is now much more efficient and able to handle large numbers of stimuli. hence, ShuffledSimpleBaselineModel is not necessary anymore and a deprecated alias to ShuffledBaselineModel diff --git a/pysaliency/baseline_utils.py b/pysaliency/baseline_utils.py index 0537a68..5794fa6 100644 --- a/pysaliency/baseline_utils.py +++ b/pysaliency/baseline_utils.py @@ -417,16 +417,6 @@ def score(self, log_bandwidth, *args, **kwargs): return val -class CrossvalGoldMultipleRegularizations(CrossvalMultipleRegularizations): - def __init__(self, stimuli, fixations, regularization_models, n_jobs=None, verbose=False): - if fixations.subject_count > 1: - crossvalidation_factory = ScikitLearnImageSubjectCrossValidationGenerator - else: - crossvalidation_factory = ScikitLearnWithinImageCrossValidationGenerator - - super().__init__(stimuli, fixations, regularization_models, crossvalidation_factory=crossvalidation_factory, n_jobs=n_jobs, verbose=verbose) - - # baseline models class GoldModel(Model):