0.1.0 - 2019-05-08¶
base¶
- Removed the
fit_predict_one
estimator method. - Removed the
fit_predict_proba_one
estimator method. - Removed the
fit_transform_one
estimator method.
compat¶
- Added
compat.convert_sklearn_to_river
. compat.convert_river_to_sklearn
now returns ansklearn.pipeline.Pipeline
when provided with acompose.Pipeline
.
compose¶
- Added
compose.Discard
. - Added
compose.Select
. - Added
compose.SplitRegressor
. - The
draw
method ofcompose.Pipeline
now works properly for arbitrary amounts of nesting, including multiple nestedcompose.FeatureUnion
.
datasets¶
- Added
datasets.fetch_electricity
.
dummy¶
- Added
dummy.NoChangeClassifier
. - Added
dummy.PriorClassifier
. - Added
dummy.StatisticRegressor
.
feature_extraction¶
- Added
feature_extraction.Differ
. - Renamed
feature_extraction.GroupBy
tofeature_extraction.Agg
. - Renamed
feature_extraction.TargetGroupBy
tofeature_extraction.TargetAgg
.
feature_selection¶
impute¶
- Added
impute.StatImputer
. - Removed
impute.CategoricalImputer
. - Removed
impute.NumericImputer
.
linear_model¶
- Added
linear_model.PAClassifier
. - Added
linear_model.PARegressor
. - Added
linear_model.SoftmaxRegression
.
metrics¶
- Added
metrics.ConfusionMatrix
. - Added
metrics.CrossEntropy
. - Added
metrics.MacroF1
. - Added
metrics.MacroPrecision
. - Added
metrics.MacroRecall
. - Added
metrics.MicroF1
. - Added
metrics.MicroPrecision
. - Added
metrics.MicroRecall
. - Each metric now has a
bigger_is_better
property to indicate if a high value is better than a low one or not.
optim¶
- Added
optim.OptimalLR
. - Added
optim.CrossEntropy
. - Removed
optim.PassiveAggressiveI
. - Removed
optim.PassiveAggressiveII
.
preprocessing¶
- Removed
preprocessing.Discarder
. - Added
on
andsparse
parameters topreprocessing.OneHotEncoder
.
stats¶
- Added
stats.Covariance
. - Added
stats.PearsonCorrelation
. - Added
stats.SmoothMean
.
utils¶
- Added
utils.check_estimator
. - Added
utils.Histogram
. - Added
utils.SortedWindow
. - Added
utils.Window
.