Scikit-learn: Ensure all attributes are documented

Created on 12 Jul 2019  ·  79Comments  ·  Source: scikit-learn/scikit-learn

as discussed in #13385 we need to ensure all attributes are documented.

if you want to work on this, you should pick a specific submodule and fix all the attribute documentation mismatches in that submodule.

Here's a script to find remaining ones (there might be some false positives):

import numpy as np
from sklearn.base import clone
from sklearn.utils.testing import all_estimators
from sklearn.utils.estimator_checks import pairwise_estimator_convert_X, enforce_estimator_tags_y
from numpydoc import docscrape

ests = all_estimators()

for name, Est in ests:
    try:
        estimator_orig = Est()
    except:
        continue
    rng = np.random.RandomState(0)
    X = pairwise_estimator_convert_X(rng.rand(40, 10), estimator_orig)
    X = X.astype(object)
    y = (X[:, 0] * 4).astype(np.int)
    est = clone(estimator_orig)
    y = enforce_estimator_tags_y(est, y)
    try:
        est.fit(X, y)
    except:
        continue
    fitted_attrs = [(x, getattr(est, x, None))
                    for x in est.__dict__.keys() if x.endswith("_")
                    and not x.startswith("_")]
    doc = docscrape.ClassDoc(type(est))
    doc_attributes = []
    incorrect = []
    for att_name, type_definition, param_doc in doc['Attributes']:
        if not type_definition.strip():
            if ':' in att_name and att_name[:att_name.index(':')][-1:].strip():
                incorrect += [name +
                              ' There was no space between the param name and '
                              'colon (%r)' % att_name]
            elif name.rstrip().endswith(':'):
                incorrect += [name +
                              ' Parameter %r has an empty type spec. '
                              'Remove the colon' % (att_name.lstrip())]

        if '*' not in att_name:
            doc_attributes.append(att_name.split(':')[0].strip('` '))
    assert incorrect == []
    fitted_attrs_names = [x[0] for x in fitted_attrs]

    bad = sorted(list(set(fitted_attrs_names) ^ set(doc_attributes)))
    if len(bad) > 0:
        msg = '{}\n'.format(name) + '\n'.join(bad)
        print("Docstring Error: Attribute mismatch in " + msg)
Documentation Easy good first issue help wanted
Source
picture amueller
🚀1 👍1

Most helpful comment

Missing attribute docstrings for each estimator

Reference this issue in your PR

  • [x] ARDRegression, [intercept_]
  • [x] AdaBoostClassifier, [base_estimator_]
  • [x] AdaBoostRegressor, [base_estimator_]
  • [x] AdditiveChi2Sampler, [sample_interval_]
  • [x] AgglomerativeClustering, [n_components_] (deprecated)
  • [x] BaggingClassifier, [n_features_]
  • [x] BaggingRegressor, [base_estimator_, n_features_]
  • [x] BayesianGaussianMixture, [mean_precision_prior, mean_precision_prior_]
  • [x] BayesianRidge, [X_offset_, X_scale_]
  • [x] BernoulliNB, [coef_, intercept_]
  • [x] BernoulliRBM, [h_samples_]
  • [ ] Birch, [fit_, partial_fit_]
  • [ ] CCA, [coef_, x_mean_, x_std_, y_mean_, y_std_]
  • [x] CheckingClassifier, [classes_]
  • [x] ComplementNB, [coef_, intercept_]
  • [x] CountVectorizer, [stop_words_, vocabulary_]
  • [ ] DecisionTreeRegressor, [classes_, n_classes_]
  • [x] DictVectorizer, [feature_names_, vocabulary_]
  • [ ] DummyClassifier, [output_2d_]
  • [ ] DummyRegressor, [output_2d_]
  • [ ] ElasticNet, [dual_gap_]
  • [ ] ElasticNetCV, [dual_gap_]
  • [ ] EllipticEnvelope, [dist_, raw_covariance_, raw_location_, raw_support_]
  • [x] ExtraTreeClassifier, [feature_importances_]
  • [ ] ExtraTreeRegressor, [classes_, feature_importances_, n_classes_]
  • [x] ExtraTreesClassifier, [base_estimator_]
  • [x] ExtraTreesRegressor, [base_estimator_]
  • [x] FactorAnalysis, [mean_]
  • [ ] FeatureAgglomeration, [n_components_]
  • [x] GaussianProcessClassifier, [base_estimator_]
  • [x] GaussianRandomProjection, [components_]
  • [x] GradientBoostingClassifier, [max_features_, n_classes_, n_features_, oob_improvement_]
  • [x] GradientBoostingRegressor, [max_features_, n_classes_, n_estimators_, n_features_, oob_improvement_]
  • [x] HistGradientBoostingClassifier, [bin_mapper_, classes_, do_early_stopping_, loss_, n_features_, scorer_]
  • [x] HistGradientBoostingRegressor, [bin_mapper_, do_early_stopping_, loss_, n_features_, scorer_]
  • [x] IncrementalPCA, [batch_size_]
  • [x] IsolationForest, [base_estimator_, estimators_features_, n_features_]
  • [x] IsotonicRegression, [X_max_, X_min_, f_]
  • [x] IterativeImputer, [random_state_]
  • [x] KNeighborsClassifier, [classes_, effective_metric_, effective_metric_params_, outputs_2d_]
  • [x] KNeighborsRegressor, [effective_metric_, effective_metric_params_]
  • [x] KernelCenterer, [K_fit_all_, K_fit_rows_]
  • [x] KernelDensity, [tree_]
  • [x] KernelPCA, [X_transformed_fit_, dual_coef_]
  • [x] LabelBinarizer, [classes_, sparse_input_, y_type_]
  • [x] LabelEncoder, [classes_]
  • [x] LarsCV, [active_]
  • [x] Lasso, [dual_gap_]
  • [x] LassoLarsCV, [active_]
  • [x] LassoLarsIC, [alphas_]
  • [x] LatentDirichletAllocation, [bound_, doc_topic_prior_, exp_dirichlet_component_, random_state_, topic_word_prior_]
  • [x] LinearDiscriminantAnalysis, [covariance_]
  • [x] LinearRegression, [rank_, singular_]
  • [x] LinearSVC, [classes_]
  • [x] LocalOutlierFactor, [effective_metric_, effective_metric_params_]
  • [x] MDS, [dissimilarity_matrix_, n_iter_]
  • [x] MLPClassifier, [best_loss_, loss_curve_, t_]
  • [x] MLPRegressor, [best_loss_, loss_curve_, t_]
  • [x] MinMaxScaler, [n_samples_seen_]
  • [x] MiniBatchDictionaryLearning, [iter_offset_]
  • [x] MiniBatchKMeans, [counts_, init_size_, n_iter_]
  • [x] MultiLabelBinarizer, [classes_]
  • [x] MultiTaskElasticNet, [dual_gap_, eps_, sparse_coef_]
  • [x] MultiTaskElasticNetCV, [dual_gap_]
  • [x] MultiTaskLasso, [dual_gap_, eps_, sparse_coef_]
  • [x] MultiTaskLassoCV, [dual_gap_]
  • [x] NearestCentroid, [classes_]
  • [x] NearestNeighbors, [effective_metric_, effective_metric_params_]
  • [x] NeighborhoodComponentsAnalysis, [random_state_]
  • [x] NuSVC, [class_weight_, fit_status_, probA_, probB_, shape_fit_]
  • [ ] NuSVR, [class_weight_, fit_status_, n_support_, probA_, probB_, shape_fit_]
  • [x] OAS, [location_]
  • [ ] OneClassSVM, [class_weight_, fit_status_, n_support_, probA_, probB_, shape_fit_]
  • [x] OneVsOneClassifier, [n_classes_]
  • [x] OneVsRestClassifier, [coef_, intercept_, n_classes_]
  • [x] OrthogonalMatchingPursuit, [n_nonzero_coefs_]
  • [ ] PLSCanonical, [coef_, x_mean_, x_std_, y_mean_, y_std_]
  • [x] PLSRegression, [x_mean_, x_std_, y_mean_, y_std_]
  • [ ] PLSSVD, [x_mean_, x_std_, y_mean_, y_std_]
  • [x] PassiveAggressiveClassifier, [loss_function_, t_]
  • [x] PassiveAggressiveRegressor, [t_]
  • [x] Perceptron, [loss_function_]
  • [x] QuadraticDiscriminantAnalysis, [classes_, covariance_]
  • [x] RBFSampler, [random_offset_, random_weights_]
  • [ ] RFE, [classes_]
  • [ ] RFECV, [classes_]
  • [x] RadiusNeighborsClassifier, [classes_, effective_metric_, effective_metric_params_, outputs_2d_]
  • [x] RadiusNeighborsRegressor, [effective_metric_, effective_metric_params_]
  • [x] RandomForestClassifier, [oob_decision_function_, oob_score_]
  • [x] RandomForestRegressor, [oob_prediction_, oob_score_]
  • [x] RandomTreesEmbedding, [base_estimator_, feature_importances_, n_features_, n_outputs_, one_hot_encoder_]
  • [x] RidgeCV, [cv_values_]
  • [x] RidgeClassifier, [classes_]
  • [x] RidgeClassifierCV, [cv_values_]
  • [x] SGDClassifier, [classes_, t_]
  • [x] SGDRegressor, [average_coef_, average_intercept_]
  • [x] SVC, [class_weight_, shape_fit_]
  • [ ] SVR, [class_weight_, fit_status_, n_support_, probA_, probB_, shape_fit_]
  • [x] SelectKBest, [pvalues_, scores_]
  • [x] ShrunkCovariance, [shrinkage]
  • [x] SkewedChi2Sampler, [random_offset_, random_weights_]
  • [x] SparseRandomProjection, [components_, density_]
  • [x] SpectralEmbedding, [n_neighbors_]
  • [x] TfidfVectorizer, [stop_words_, vocabulary_]
picture thomasjpfan on 13 Jul 2019
👍6

All 79 comments

I have already found at least one mismatch in attribute documentation in NMF class description. I think I can take some of this work. I am almost ready to propose some changes within decomposition and random_projection submodules.

alexitkes picture alexitkes on 12 Jul 2019

Missing attribute docstrings for each estimator

Reference this issue in your PR

  • [x] ARDRegression, [intercept_]
  • [x] AdaBoostClassifier, [base_estimator_]
  • [x] AdaBoostRegressor, [base_estimator_]
  • [x] AdditiveChi2Sampler, [sample_interval_]
  • [x] AgglomerativeClustering, [n_components_] (deprecated)
  • [x] BaggingClassifier, [n_features_]
  • [x] BaggingRegressor, [base_estimator_, n_features_]
  • [x] BayesianGaussianMixture, [mean_precision_prior, mean_precision_prior_]
  • [x] BayesianRidge, [X_offset_, X_scale_]
  • [x] BernoulliNB, [coef_, intercept_]
  • [x] BernoulliRBM, [h_samples_]
  • [ ] Birch, [fit_, partial_fit_]
  • [ ] CCA, [coef_, x_mean_, x_std_, y_mean_, y_std_]
  • [x] CheckingClassifier, [classes_]
  • [x] ComplementNB, [coef_, intercept_]
  • [x] CountVectorizer, [stop_words_, vocabulary_]
  • [ ] DecisionTreeRegressor, [classes_, n_classes_]
  • [x] DictVectorizer, [feature_names_, vocabulary_]
  • [ ] DummyClassifier, [output_2d_]
  • [ ] DummyRegressor, [output_2d_]
  • [ ] ElasticNet, [dual_gap_]
  • [ ] ElasticNetCV, [dual_gap_]
  • [ ] EllipticEnvelope, [dist_, raw_covariance_, raw_location_, raw_support_]
  • [x] ExtraTreeClassifier, [feature_importances_]
  • [ ] ExtraTreeRegressor, [classes_, feature_importances_, n_classes_]
  • [x] ExtraTreesClassifier, [base_estimator_]
  • [x] ExtraTreesRegressor, [base_estimator_]
  • [x] FactorAnalysis, [mean_]
  • [ ] FeatureAgglomeration, [n_components_]
  • [x] GaussianProcessClassifier, [base_estimator_]
  • [x] GaussianRandomProjection, [components_]
  • [x] GradientBoostingClassifier, [max_features_, n_classes_, n_features_, oob_improvement_]
  • [x] GradientBoostingRegressor, [max_features_, n_classes_, n_estimators_, n_features_, oob_improvement_]
  • [x] HistGradientBoostingClassifier, [bin_mapper_, classes_, do_early_stopping_, loss_, n_features_, scorer_]
  • [x] HistGradientBoostingRegressor, [bin_mapper_, do_early_stopping_, loss_, n_features_, scorer_]
  • [x] IncrementalPCA, [batch_size_]
  • [x] IsolationForest, [base_estimator_, estimators_features_, n_features_]
  • [x] IsotonicRegression, [X_max_, X_min_, f_]
  • [x] IterativeImputer, [random_state_]
  • [x] KNeighborsClassifier, [classes_, effective_metric_, effective_metric_params_, outputs_2d_]
  • [x] KNeighborsRegressor, [effective_metric_, effective_metric_params_]
  • [x] KernelCenterer, [K_fit_all_, K_fit_rows_]
  • [x] KernelDensity, [tree_]
  • [x] KernelPCA, [X_transformed_fit_, dual_coef_]
  • [x] LabelBinarizer, [classes_, sparse_input_, y_type_]
  • [x] LabelEncoder, [classes_]
  • [x] LarsCV, [active_]
  • [x] Lasso, [dual_gap_]
  • [x] LassoLarsCV, [active_]
  • [x] LassoLarsIC, [alphas_]
  • [x] LatentDirichletAllocation, [bound_, doc_topic_prior_, exp_dirichlet_component_, random_state_, topic_word_prior_]
  • [x] LinearDiscriminantAnalysis, [covariance_]
  • [x] LinearRegression, [rank_, singular_]
  • [x] LinearSVC, [classes_]
  • [x] LocalOutlierFactor, [effective_metric_, effective_metric_params_]
  • [x] MDS, [dissimilarity_matrix_, n_iter_]
  • [x] MLPClassifier, [best_loss_, loss_curve_, t_]
  • [x] MLPRegressor, [best_loss_, loss_curve_, t_]
  • [x] MinMaxScaler, [n_samples_seen_]
  • [x] MiniBatchDictionaryLearning, [iter_offset_]
  • [x] MiniBatchKMeans, [counts_, init_size_, n_iter_]
  • [x] MultiLabelBinarizer, [classes_]
  • [x] MultiTaskElasticNet, [dual_gap_, eps_, sparse_coef_]
  • [x] MultiTaskElasticNetCV, [dual_gap_]
  • [x] MultiTaskLasso, [dual_gap_, eps_, sparse_coef_]
  • [x] MultiTaskLassoCV, [dual_gap_]
  • [x] NearestCentroid, [classes_]
  • [x] NearestNeighbors, [effective_metric_, effective_metric_params_]
  • [x] NeighborhoodComponentsAnalysis, [random_state_]
  • [x] NuSVC, [class_weight_, fit_status_, probA_, probB_, shape_fit_]
  • [ ] NuSVR, [class_weight_, fit_status_, n_support_, probA_, probB_, shape_fit_]
  • [x] OAS, [location_]
  • [ ] OneClassSVM, [class_weight_, fit_status_, n_support_, probA_, probB_, shape_fit_]
  • [x] OneVsOneClassifier, [n_classes_]
  • [x] OneVsRestClassifier, [coef_, intercept_, n_classes_]
  • [x] OrthogonalMatchingPursuit, [n_nonzero_coefs_]
  • [ ] PLSCanonical, [coef_, x_mean_, x_std_, y_mean_, y_std_]
  • [x] PLSRegression, [x_mean_, x_std_, y_mean_, y_std_]
  • [ ] PLSSVD, [x_mean_, x_std_, y_mean_, y_std_]
  • [x] PassiveAggressiveClassifier, [loss_function_, t_]
  • [x] PassiveAggressiveRegressor, [t_]
  • [x] Perceptron, [loss_function_]
  • [x] QuadraticDiscriminantAnalysis, [classes_, covariance_]
  • [x] RBFSampler, [random_offset_, random_weights_]
  • [ ] RFE, [classes_]
  • [ ] RFECV, [classes_]
  • [x] RadiusNeighborsClassifier, [classes_, effective_metric_, effective_metric_params_, outputs_2d_]
  • [x] RadiusNeighborsRegressor, [effective_metric_, effective_metric_params_]
  • [x] RandomForestClassifier, [oob_decision_function_, oob_score_]
  • [x] RandomForestRegressor, [oob_prediction_, oob_score_]
  • [x] RandomTreesEmbedding, [base_estimator_, feature_importances_, n_features_, n_outputs_, one_hot_encoder_]
  • [x] RidgeCV, [cv_values_]
  • [x] RidgeClassifier, [classes_]
  • [x] RidgeClassifierCV, [cv_values_]
  • [x] SGDClassifier, [classes_, t_]
  • [x] SGDRegressor, [average_coef_, average_intercept_]
  • [x] SVC, [class_weight_, shape_fit_]
  • [ ] SVR, [class_weight_, fit_status_, n_support_, probA_, probB_, shape_fit_]
  • [x] SelectKBest, [pvalues_, scores_]
  • [x] ShrunkCovariance, [shrinkage]
  • [x] SkewedChi2Sampler, [random_offset_, random_weights_]
  • [x] SparseRandomProjection, [components_, density_]
  • [x] SpectralEmbedding, [n_neighbors_]
  • [x] TfidfVectorizer, [stop_words_, vocabulary_]
thomasjpfan picture thomasjpfan on 13 Jul 2019
👍6

I can take up the tree submodule attribute documentation mismatches, which includes:

  • DecisionTreeRegressor, [classes_, n_classes_]
  • ExtraTreeClassifier, [classes_, max_features_, n_classes_, n_features_, n_outputs_, tree_]
  • ExtraTreeRegressor, [classes_, max_features_, n_classes_, n_features_, n_outputs_, tree_]
mepa picture mepa on 13 Jul 2019

I'm working on LinearRegression, [rank_, singular_].

wendyhhu picture wendyhhu on 13 Jul 2019
👍1

I'm working on LinearSVC, [n_iter_] and LinearSVR, [n_iter_]

wendyhhu picture wendyhhu on 13 Jul 2019
👍1

I'll take up Gradient boosting i.e.

  • GradientBoostingClassifier [base_estimator_, max_features_, n_classes_, n_features_]

    • GradientBoostingRegressor [base_estimator_, classes_, max_features_, n_estimators_, n_features_]

matsmaiwald picture matsmaiwald on 14 Jul 2019

nevermind, misread where attributes are missing and where not

matsmaiwald picture matsmaiwald on 14 Jul 2019

It's looking like there is also classes_ attribute undocumented for classifiers of naive_bayes submodule. I have started to fix it.

alexitkes picture alexitkes on 14 Jul 2019

I will work on TfidfVectorizer, [fixed_vocabulary_]

mandalbiswadip picture mandalbiswadip on 14 Jul 2019

I will work on:

  • RandomForestClassifier, [base_estimator_]
  • RandomForestRegressor, [base_estimator_, n_classes_]
  • ExtraTreesClassifier, [base_estimator_]
  • ExtraTreesRegressor, [base_estimator_, n_classes_]
rcwoolston picture rcwoolston on 14 Jul 2019

I'm working on:

  • SGDClassifier, [average_coef_, average_intercept_, standard_coef_, standard_intercept_]
  • SGDRegressor, [standard_coef_, standard_intercept_]

EDIT: opened an issue to change these attributes from public to private (reference: #14364)

wendyhhu picture wendyhhu on 14 Jul 2019

I am working on:
KernelCenterer, [K_fit_all_, K_fit_rows_]
MinMaxScaler, [n_samples_seen_]

SwordKnight6216 picture SwordKnight6216 on 14 Jul 2019

I will work on:

  • RandomTreesEmbedding, [base_estimator_, classes_, feature_importances_, n_classes_, n_features_, n_outputs_, one_hot_encoder_]
rcwoolston picture rcwoolston on 14 Jul 2019

I also have discovered the KNeighborsClassifier, KNeighborsRegressor and possibly other classes of neighbors module have no attribute documentation at all. Currently working on KNeighborsRegressor that has 2 attributes:

  • effective_metric_
  • effective_metric_params_

The KNeighborsClassifier class has four attributes:

  • classes_
  • effective_metric_
  • effective_metric_params_
  • outputs_2d_
alexitkes picture alexitkes on 14 Jul 2019

@alexitkes good catch. Thanks!

amueller picture amueller on 14 Jul 2019

Working on QuadraticDiscriminantAnalysis, [classes_, covariance_]

abhishek-jana picture abhishek-jana on 15 Jul 2019

Working on KNeighborsClassifier, [classes_, effective_metric_, effective_metric_params_, outputs_2d_]
RadiusNeighborsClassifier, [classes_, effective_metric_, effective_metric_params_, outputs_2d_]

abhishek-jana picture abhishek-jana on 15 Jul 2019

Working on:
LinearSVC, [classes_]
NuSVC, [class_weight_, classes_, fit_status_, probA_, probB_, shape_fit_]
SVC, [class_weight_, classes_, shape_fit_]

kwinata picture kwinata on 20 Jul 2019

Working on:

  • [ ] BaggingClassifier, [n_features_, oob_decision_function_, oob_score_]
  • [ ] BaggingRegressor, [base_estimator_, n_features_, oob_prediction_, oob_score_]
  • [ ] AdaBoostClassifier, [base_estimator_]
  • [ ] AdaBoostRegressor, [base_estimator_]
mdomarsaleem picture mdomarsaleem on 25 Jul 2019

Working on:

CountVectorizer, [stop_words_, vocabulary_]
DictVectorizer, [feature_names_, vocabulary_]

pvsagar picture pvsagar on 25 Jul 2019

Hii!! I'd like to help out with this one.. Can anyone plz tell me where should I start??

ManishAradwad picture ManishAradwad on 5 Aug 2019

We are working on the functions in dict_learning.py @spbail

hannahbrucemacdonald picture hannahbrucemacdonald on 24 Aug 2019

Working on LinearDiscriminantAnalysis with @olgadk7

m-clare picture m-clare on 24 Aug 2019

Working on Attribute mismatch in RidgeClassifierCV @npatta01

ingrid88 picture ingrid88 on 24 Aug 2019

Working on DecisionTreeRegressor with @ingrid88 + @npatta01

meiguan picture meiguan on 24 Aug 2019

Working on LinearDiscriminantAnalysis with @olgadk7

False positive for the attribute script above. This has been documented.

m-clare picture m-clare on 24 Aug 2019

Working on AdditiveChi2Sampler with @olgadk7

m-clare picture m-clare on 24 Aug 2019

Working on LabelEncoder with @eugeniaft

FranciDona picture FranciDona on 24 Aug 2019

will try to work on randomtreeclassifier!

joyharjanto picture joyharjanto on 24 Aug 2019

working on

Perceptron

joanaz picture joanaz on 24 Aug 2019

working on BernoulliRBM

npatta01 picture npatta01 on 24 Aug 2019

Working on ExtraTreeClassifer

she-dares picture she-dares on 24 Aug 2019

Working on LabelEncoder with @eugeniaft

LabelEncoder looks like it has no mismatch, we're working on OneClassSVM

FranciDona picture FranciDona on 24 Aug 2019

I think the tree regressors should deprecate their classes instead.

amueller picture amueller on 24 Aug 2019

working on SVR

ingrid88 picture ingrid88 on 24 Aug 2019

Working on:

  • OneVsOneClassifier, [n_classes_]
  • OneVsRestClassifier, [coef_, intercept_, n_classes_]
YuliaZamriy picture YuliaZamriy on 24 Aug 2019

working on LinearRegression, [rank_, singular_]

arzoobh picture arzoobh on 24 Aug 2019

working on LatentDirichletAllocation, [bound_, doc_topic_prior_, exp_dirichlet_component_, random_state_, topic_word_prior_]

YuliaZamriy picture YuliaZamriy on 24 Aug 2019

working on
BaggingClassifier, [n_features_, oob_decision_function_, oob_score_]
BaggingRegressor, [base_estimator_, n_features_, oob_prediction_, oob_score_]

joanaz picture joanaz on 24 Aug 2019

BaggingClassifier, [n_features_, oob_decision_function_, oob_score_]
BaggingRegressor, [base_estimator_, n_features_, oob_prediction_, oob_score_]
oob_ attributes are address in PR #14779, n_features_ & base_estimator_ are false positives.

joanaz picture joanaz on 24 Aug 2019

working on
AdaBoostClassifier, [base_estimator_]

Update: was already fixed in https://github.com/scikit-learn/scikit-learn/pull/14477

nitya picture nitya on 24 Aug 2019

I think we should not recommend this issue for the next sprints, or use a much more curated version.

Based on my experience on the previous sprint, there are still a lot of false positives, and we end up asking contributors to actually deprecate public attributes to make them private, which is arguably much harder (and can be frustrating since contributors feel they worked for nothing).

Ping @amueller @thomasjpfan WDYT?

NicolasHug picture NicolasHug on 3 Sep 2019

I think we should not recommend this issue for the next sprints, or use a much more curated version.

Maybe if we had a general validation tool for docstring such as proposed in https://github.com/numpy/numpydoc/issues/213 things would be a bit easier for contributors. Although I agree that it doesn't fully address the fact that some attributes are public while they shouldn't be.

rth picture rth on 7 Oct 2019

TfidfVectorizer, SpectralEmbedding, SparseRandomProjection are updated.

zioalex picture zioalex on 12 Oct 2019

I was wondering about taking that issue as my first one, but after some random picking of submodules listed by script, the only classes I found as uncorrectly documented are PLS* classes. But they live in _pls_.py file, which seems to be non-public. Should I work on them or find another good first issue?

panpiort8 picture panpiort8 on 24 Oct 2019

As long as the actual classes are public, they qualify. The public classes are listed in doc/modules/classes.rst. The PLS* classes are there so feel free to document them

NicolasHug picture NicolasHug on 24 Oct 2019

Does it make sense to alphabetize all of the attributes as well? I think it would provide structure to the section and make the section easier to read.

pwalchessen picture pwalchessen on 2 Nov 2019
👍1

@pwalchessen I agree, sounds like a good idea. As mentioned in person, I would also add that to the test.

amueller picture amueller on 2 Nov 2019

These seem still open and kinda obvious:

Docstring Error: Attribute mismatch in RidgeCV
cv_values_
Docstring Error: Attribute mismatch in RidgeClassifier
classes_
Docstring Error: Attribute mismatch in RidgeClassifierCV
classes_
cv_values_
Docstring Error: Attribute mismatch in SkewedChi2Sampler
random_offset_
random_weights_
Docstring Error: Attribute mismatch in PLSCanonical
coef_
x_mean_
x_std_
y_mean_
y_std_
Docstring Error: Attribute mismatch in PLSRegression
x_mean_
x_std_
y_mean_
y_std_
Docstring Error: Attribute mismatch in PLSSVD
x_mean_
x_std_
y_mean_
y_std_
Docstring Error: Attribute mismatch in PassiveAggressiveClassifier
loss_function_
Docstring Error: Attribute mismatch in Perceptron
loss_function_
Docstring Error: Attribute mismatch in PolynomialFeatures
powers_
Docstring Error: Attribute mismatch in QuadraticDiscriminantAnalysis
covariance_
Docstring Error: Attribute mismatch in RBFSampler
random_offset_
random_weights_
Docstring Error: Attribute mismatch in RadiusNeighborsClassifier
n_samples_fit_
outlier_label_
Docstring Error: Attribute mismatch in RadiusNeighborsRegressor
n_samples_fit_
Docstring Error: Attribute mismatch in RadiusNeighborsTransformer
effective_metric_
effective_metric_params_
n_samples_fit_
Docstring Error: Attribute mismatch in ElasticNet
dual_gap_
sparse_coef_
Docstring Error: Attribute mismatch in ElasticNetCV
dual_gap_
Docstring Error: Attribute mismatch in EllipticEnvelope
dist_
raw_covariance_
raw_location_
raw_support_

and a bunch more...

amueller picture amueller on 2 Nov 2019

Updated list of outstanding attributes that need to be added.

  • [ ] BayesianGaussianMixture

    • [x] mean_precision_prior

    • [ ] mean_precision_prior_

  • [ ] BayesianRidge

    • [ ] X_offset_

    • [ ] X_scale_

  • [ ] BernoulliNB

    • [ ] coef_ array

    • [ ] intercept_

  • [ ] Birch

    • [ ] fit_

    • [ ] partial_fit_

  • [ ] CCA

    • [ ] coef_ array, shape (1, n_features) or (n_classes, n_features); Coefficient of the features in the decision function.

    • [ ] x_mean_ :  array, shape (n_features, ) The mean over features.

    • [ ] x_std_

    • [ ] y_mean_

    • [ ] y_std_

  • [x] CategoricalNB

    • [x] classes_ (classes_ : array, shape (n_classes, )

      A list of class labels known to the classifier.

  • [ ] ComplementNB

    • [ ] coef_: array, shape (1, n_features) or (n_classes, n_features); Coefficient of the features in the decision function.

    • [ ] intercept_

  • [x] CountVectorizer

    • [x] stop_words_

    • [x] vocabulary_

  • [x] DecisionTreeClassifier

    • [x] feature_importances_

  • [ ] DecisionTreeRegressor

    • [ ] classes_ : array-like, shape (n_classes,); Unique class labels

    • [ ] n_classes_ : int; Number of unique class labels

    • [x] feature_importances_

  • [ ] DictVectorizer

    • [ ] feature_names_

    • [ ] vocabulary_

  • [ ] DummyClassifier

    • [ ] output_2d_

  • [ ] DummyRegressor

    • [ ] output_2d_

  • [ ] ElasticNet

    • [ ] dual_gap_

    • [ ] sparse_coef_

  • [ ] ElasticNetCV

    • [ ] dual_gap_

  • [ ] EllipticEnvelope

    • [ ] dist_

    • [ ] raw_covariance_

    • [ ] raw_location_

    • [ ] raw_support_

  • [ ] ExtraTreeClassifier

    • [ ] feature_importances_

  • [ ] ExtraTreeRegressor

    • [ ] classes_ : array-like, shape (n_classes,); Unique class labels

    • [ ] feature_importances_

    • [ ] n_classes_ : int; Number of unique class labels

  • [ ] FeatureAgglomeration

    • [ ] n_components_

    • [x] distances_

  • [ ] GaussianProcessClassifier

    • [ ] base_estimator_

    • [x] kernel_

  • [x] GaussianRandomProjection

    • [x] components_

  • [ ] GradientBoostingClassifier

    • [ ] max_features_

    • [ ] n_classes_ : int; Number of unique classes.

    • [ ] n_features_ : int; Number of features used.

    • [x] oob_improvement_

    • [x] feature_importances_

  • [ ] GradientBoostingRegressor

    • [ ] max_features_

    • [ ] n_classes_ : int; Number of unique classes.

    • [ ] n_estimators_

    • [ ] n_features_ : int; Number of features used.

    • [x] oob_improvement_

    • [x] feature_importances_

  • [ ] HistGradientBoostingClassifier

    • [ ] bin_mapper_

    • [ ] classes_

    • [ ] do_early_stopping_

    • [ ] loss_

    • [ ] n_features_ : int; The number of selected features.

    • [x] n_iter_

    • [ ] scorer_

  • [ ] HistGradientBoostingRegressor

    • [ ] bin_mapper_

    • [ ] do_early_stopping_

    • [ ] loss_

    • [ ] n_features_ : int; The number of selected features.

    • [ ]

    • [ ] scorer_

  • [ ] IncrementalPCA

    • [ ] batch_size_

  • [ ] IsolationForest

    • [ ] base_estimator_

    • [ ] estimators_features_

    • [x] estimators_samples_

    • [ ] n_features_ : int; The number of selected features.

  • [ ] KernelCenterer

    • [ ] K_fit_all_

    • [ ] K_fit_rows_

  • [ ] KernelDensity

    • [ ] tree_

  • [ ] LarsCV

    • [ ] active_

  • [ ] Lasso

    • [ ] dual_gap_

    • [x] sparse_coef_

  • [ ] LassoLarsCV

    • [ ] active_

  • [ ] LassoLarsIC

    • [ ] alphas_

  • [ ] LatentDirichletAllocation

    • [x] bound_

    • [x] doc_topic_prior_

    • [ ] exp_dirichlet_component_

    • [ ] random_state_

  • [ ] LocalOutlierFactor

    • [ ] effective_metric_

    • [ ] effective_metric_params_

    • [ ] n_samples_fit_ : int; Number of samples in the fitted data.

  • [ ] MDS

    • [ ] dissimilarity_matrix_

    • [ ] n_iter_ : int; Number of iterations. 

  • [ ] MLPClassifier

    • [ ] best_loss_

    • [ ] loss_curve_

    • [ ] t_

  • [ ] MLPRegressor

    • [ ] best_loss_

    • [ ] loss_curve_

    • [ ] t_

  • [ ] MiniBatchKMeans

    • [ ] counts_

    • [ ] init_size_

    • [ ] n_iter_: int; Number of iterations. 

  • [ ] MultiTaskElasticNet

    • [ ] dual_gap_

    • [ ] eps_

    • [ ] sparse_coef_

  • [ ] MultiTaskElasticNetCV

    • [ ] dual_gap_

  • [ ] MultiTaskLasso

    • [ ] dual_gap_

    • [ ] eps_

    • [ ] sparse_coef_

  • [ ] MultiTaskLassoCV

    • [ ] dual_gap_

  • [ ] OAS

    • [ ] location_

  • [ ] OneVsRestClassifier

    • [ ] coef_ : array, shape (1, n_features) or (n_classes, n_features); Coefficient of the features in the decision function.

    • [ ] intercept_

    • [ ] n_classes_ : int; Number of unique classes.

  • [ ] OrthogonalMatchingPursuit

    • [ ] n_nonzero_coefs_

  • [ ] PLSCanonical

    • [ ] coef_ : array, shape (1, n_features) or (n_classes, n_features); Coefficient of the features in the decision function.

    • [ ] x_mean_ : float???; Mean of

    • [ ] x_std_

    • [ ] y_mean_

    • [ ] y_std_

  • [ ] PLSRegression

    • [ ] x_mean_

    • [ ] x_std_

    • [ ] y_mean_

    • [ ] y_std_

  • [ ] PLSSVD

    • [ ] x_mean_

    • [ ] x_std_

    • [ ] y_mean_

    • [ ] y_std_

  • [ ] PassiveAggressiveClassifier

    • [ ] loss_function_

  • [ ] RBFSampler

    • [ ] random_offset_

    • [ ] random_weights_

  • [ ] ShrunkCovariance

    • [ ] shrinkage

  • [ ] SkewedChi2Sampler

    • [ ] random_offset_

    • [ ] random_weights_

  • [ ] _BaseRidgeCV

    • [ ] alpha_

    • [ ] coef_

    • [ ] intercept_

  • [ ] _ConstantPredictor

    • [ ] y_

  • [ ] _RidgeGCV

    • [ ] alpha_

    • [ ] coef_

    • [ ] dual_coef_

    • [ ] intercept_

jigna-panchal picture jigna-panchal on 3 Nov 2019

I am going to add feature_importances_ to the documentation for ExtraTreeRegressor

marielledado picture marielledado on 25 Jan 2020

A group of data science majors and I will begin working on the BayesianRidge, [X_offset_, X_scale_] attribute documentation.

juliapiscioniere picture juliapiscioniere on 28 Jan 2020

Hi, our group of contributors will be working on:

  • PLSSVD
  • CCA
  • Incremental PCA
  • MiniBatchKMeans
  • Lasso
phork37 picture phork37 on 20 Feb 2020

Potential fixes in #16826

BenjaminLiuPenrose picture BenjaminLiuPenrose on 2 Apr 2020

The test was added in #16286.
There are currently still a couple of classes that are skipped:
https://github.com/scikit-learn/scikit-learn/blob/753da1de06a764f264c3f5f4817c9190dbe5e021/sklearn/tests/test_docstring_parameters.py#L180

Some of these already have PRs, so make sure to check that before starting to work on it.

amueller picture amueller on 29 May 2020

Some of these already have PRs, so make sure to check that before starting to work on it.

A good option would also be to try to look at open PRs that have not been merged and try to finish them.

As a rule of thumb, if a PR hasn't got some activity for more than 2-3 weeks, it is fine to try to take it over and try to finish it.

lesteve picture lesteve on 23 Jun 2020

In case your are interested in such solution, there is a way to implement an extension for sphinx which checks that parameters are all documented or not mispelled (you can see an example here: https://github.com/sdpython/pyquickhelper/blob/master/src/pyquickhelper/sphinxext/sphinx_docassert_extension.py). Maybe it can be useful to add a custom one to scikit-learn documentation.

sdpython picture sdpython on 23 Jun 2020
🎉1

@sdpython, that would be wonderful! If you are not working on something else, perhaps you could propose a draft PR? Thanks!

cmarmo picture cmarmo on 23 Jun 2020

Interesting!

IIRC we have a common tests that check that all attributes are documented. It was added in https://github.com/scikit-learn/scikit-learn/pull/16286. Also I seem to remember that mne-python had something similar.

I don't have an informed opinion on which approach is preferable but I would say that documenting the missing parameters is probably higher priority than deciding how we want to do the checking.

lesteve picture lesteve on 23 Jun 2020

The issue with doing that in sphinx that in our case building a documentation takes a long time (due to generating all the examples) so a unit test or standalone tool would be easier to use. Note that we have previously used numpydoc validation in https://github.com/scikit-learn/scikit-learn/issues/15440 and some validation of the docstring with type annotations could be done with https://github.com/terrencepreilly/darglint. So we probably should avoid a situation of using 5 different validation tools for docstrings as well :)

rth picture rth on 23 Jun 2020

I like the ability to use pytest to check the results, for instance:

pytest -v  --runxfail -k IsolationForest sklearn/tests/test_docstring_parameters.py

so maybe it's not necessary to change our sphinx build for this.

ogrisel picture ogrisel on 23 Jun 2020

I checked which attribute docstrings are still missing (the list above is out of date). These are the ones I found:

BayesianGaussianMixture, [mean_precision_prior]
BayesianRidge, [X_offset_, X_scale_]
BernoulliNB, [coef_, intercept_]
Birch, [fit_, partial_fit_]
CCA, [x_mean_, x_std_, y_mean_, y_std_]
DecisionTreeRegressor, [classes_, n_classes_]
DummyClassifier, [output_2d_]
DummyRegressor, [output_2d_]
ElasticNet, [dual_gap_]
ElasticNetCV, [dual_gap_]
ExtraTreeRegressor, [classes_, n_classes_]
FeatureAgglomeration, [n_components_]
LarsCV, [active_]
Lasso, [dual_gap_]
LassoLarsCV, [active_]
LassoLarsIC, [alphas_]
MiniBatchKMeans, [counts_, init_size_, n_iter_]
MultiTaskElasticNet, [dual_gap_, eps_, sparse_coef_]
MultiTaskElasticNetCV, [dual_gap_]
MultiTaskLasso, [dual_gap_, eps_, sparse_coef_]
MultiTaskLassoCV, [dual_gap_]
NuSVR, [probA_, probB_]
OneClassSVM, [probA_, probB_]
OneVsRestClassifier, [coef_, intercept_]
OrthogonalMatchingPursuit, [n_nonzero_coefs_]
PLSCanonical, [x_mean_, x_std_, y_mean_, y_std_]
PLSSVD, [x_mean_, x_std_, y_mean_, y_std_]
SVR, [probA_, probB_]

marenwestermann picture marenwestermann on 26 Jun 2020

Thanks @marenwestermann!

cmarmo picture cmarmo on 26 Jun 2020

I'm working on MiniBatchKMeans

jeremiedbb picture jeremiedbb on 8 Jul 2020

I'm working on Lasso.

marenwestermann picture marenwestermann on 10 Jul 2020

I'm now working on adding the attribute sparse_coef_ to MultiTaskElasticNet and MultiTaskLasso.

marenwestermann picture marenwestermann on 10 Jul 2020

I'm working on LarsCV.

marenwestermann picture marenwestermann on 10 Jul 2020

@thomasjpfan it is said in the classes SVR and OneClassSVM:
"The probA_ attribute is deprecated in version 0.23 and will be removed in version 0.25." and
"The probB_ attribute is deprecated in version 0.23 and will be removed in version 0.25."

Therefore, these attributes probably don't need documentation anymore, right?
Going from here, will these two attributes also be deprecated in the class NuSVR?

marenwestermann picture marenwestermann on 7 Aug 2020

The attributes classes_ and n_classes_ for ExtraTreeRegressor are false positives.

marenwestermann picture marenwestermann on 11 Sep 2020

Therefore, these attributes probably don't need documentation anymore, right?
Going from here, will these two attributes also be deprecated in the class NuSVR?

Since we are deprecating them I would say we would not need document them.

The attributes classes_ and n_classes_ for ExtraTreeRegressor are false positives.

Yup those should be deprecated then removed if they are not already.

thomasjpfan picture thomasjpfan on 11 Sep 2020

The DecisionTreeRegressor class says:
"the n_classes_ attribute is to be deprecated from version 0.22 and will be removed in 0.24."
"the classes_ attribute is to be deprecated from version 0.22 and will be removed in 0.24."

So these attributes also don't need documentation right?

Abilityguy picture Abilityguy on 16 Sep 2020

So these attributes also don't need documentation right?

Right @Abilityguy, thanks for pointing out that.

cmarmo picture cmarmo on 16 Sep 2020

I can see below mismatch in _RidgeGCV :
Docstring Error: Attribute mismatch in _RidgeGCV
alpha_
best_score_
coef_
dual_coef_
intercept_
n_features_in_

and in _BaseRidgeCV:
Docstring Error: Attribute mismatch in _BaseRidgeCV
alpha_
best_score_
coef_
intercept_
n_features_in_

Can I take it up? I am first timer and wants to contribute.

mynkdsi1011 picture mynkdsi1011 on 24 Sep 2020

@marenwestermann in the class FeatureAgglomeration, it is said that, in version 0.21, n_connected_components_ was added to replace n_components_, then n_components_ would be false positive right..?

srivathsa729 picture srivathsa729 on 26 Sep 2020

@srivathsa729 from my understanding yes. However, it would be good if one of the core developers could double check.

marenwestermann picture marenwestermann on 29 Sep 2020

I will take up ElasticNet

disha4u picture disha4u on 5 Oct 2020

Documentation of the attributes X_offset_ and X_scale_ for BayesianRidge has been added with #18607 .

marenwestermann picture marenwestermann on 4 Nov 2020
👍1

The attribute output_2d_ is deprecated in DummyClassifier and DummyRegressor (see #14933).

marenwestermann picture marenwestermann on 5 Nov 2020

I ran the script provided by @amueller at the top of this PR (the code needs to be slightly modified because things have moved around). I couldn't find any more attributes that need to be documented with the exception of n_features_in_ which I see has been introduced in #16112. This attribute is undocumented in I think all classes it was introduced to. Should it be documented?
ping @NicolasHug

marenwestermann picture marenwestermann on 9 Nov 2020
Was this page helpful?
0 / 5 - 0 ratings

Related issues

vitorcoliveira picture vitorcoliveira  ·  3Comments
yinruiqing picture yinruiqing  ·  3Comments
trchan picture trchan  ·  3Comments
amueller picture amueller  ·  3Comments
AntiDoctor picture AntiDoctor  ·  3Comments