Skip to content

Latest commit

 

History

History
216 lines (155 loc) · 12.5 KB

ESTIMATOR_OVERVIEW.md

File metadata and controls

216 lines (155 loc) · 12.5 KB
Raw

Overview of sktime's estimators

Table of contents

Transformers (simple)

Simple (or first-degree) transformations:

Atoms

Single time series to primitives

Name Class Maintainer References
e.g. Fitted parameter feature extraction

Single time series to single time series

Name Class Maintainer References
e.g. Fourier transform

Nested data frame to nested data frame

Name Class Maintainer References
Interval segmenter (fixed) transformers.compose.IntervalSegmenter @mloning
Interval segmenter (random) transformers.compose.RandomIntervalSegmenter @mloning
Piecewise Aggregate Approximation transformers.panel.dictionary_based._paa.PAA @MatthewMiddlehurst Keogh et al (2001) - Dimensionality reduction for fast similarity search in large time series databases
Symbolic Aggregate Approximation transformers.panel.dictionary_based._sax.SAX @MatthewMiddlehurst Lin et al (2007) - Experiencing SAX: a novel symbolic representation of time series
Symbolic Fourier Approximation transformers.panel.dictionary_based._sfa.SFA @MatthewMiddlehurst Schäfer (2012) - SFA: a symbolic fourier approximation and index for similarity

Nested data frame to tabular data frame

Name Class Maintainer References
Tabularise (UK) transformers.compose.Tabulariser @mloning
Tabularize (US) transformers.compose.Tabularizer @mloning
Auto-correlation function transformers.spectral_based.AutoCorrelationFourierTransformer @jsellier
Cosine Transform transformers.spectral_based.CosineTransformer @jsellier
Discrete Fourier Transform transformers.spectral_based.DiscreteFourierTransformer @jsellier
Power Spectrum transformers.spectral_based.PowerSpectrumTransformer @jsellier
tsfresh Feature Extractor transformers.summarise._tsfresh.TSFreshFeatureExtractor @mloning @Ayushmaanseth
tsfresh Relevant Feature Extractor transformers.summarise._tsfresh.TSFreshRelevantFeatureExtractor @mloning @Ayushmaanseth
Derivative Series transformers.summarise.DeriativeSlopeTransformer @mloning
Plateau Finder transformers.summarise.PlateauFinder @mloning
Random Interval Feature Extractor transformers.summarise.RandomIntervalFeatureExtractor @mloning
Matrix profile transformers.matrix_profile Claudia Rincon Sanchez (custom implementation)
Principal component scores after tabularization transformers.PCATransformer @prockenschaub Hotelling (1933) - Analysis of a complex of statistical variables into principal components
Shapelet transform transformers.ShapeletTransform @jasonlines Hills et al (2014) - Classification of time series by shapelet transformation
Shapelet transform (contracted) transformers.ContractedShapeletTransform @jasonlines Hills et al (2014) - Classification of time series by shapelet transformation
Shapelet transform (random sampled) transformers.RandomEnumerationShapeletTransform @jasonlines Hills et al (2014) - Classification of time series by shapelet transformation
ROCKET transformers.rocket.Rocket @angus924 Dempser et al (2019) ROCKET: Exceptionally fast and accurate time series classification using random convolutional kernels
Canonical Time-series Characteristics contrib.transformers.catch22_features.Catch22 @MatthewMiddlehurst Lubba et al (2019) - catch22: CAnonical Time-series CHaracteristics

Multivariate nested data frame to univariate nested data frame (n-mts-to-n-1-ts)

Name Class Maintainer References
Concatenate variables transformers.compose.ColumnConcatenator @mloning

Composition

Pipeline

| Name | Class | Maintainer | References | | ------| ------ | ------- | ------ | ------- | | n-ts-to-X | Concatenate column-wise | transformers.compose.ColumnTransformer | @mloning | | | n-ts-to-X | Feature union | pipeline.FeatureUnion | @mloning | |

Reduction

From/output To/input Name Class Maintainer References
n-ts-to-df 1-ts-to-df Apply row-wise transformers.compose.RowwiseTransformer @mloning

Transformers (paired)

Paired (or second-degree) transformations:

Note: the interface for 2nd degree transformers is currently under re-factoring, and currently not consistent or homogenous.

Atoms

Distances

Name Class Maintainer References
BOSS Distance classification.dictionary_based._boss.boss_distance @MatthewMiddlehurst Schäfer (2014) - The BOSS is concerned with time series classification in the presence of noise
Histogram Intersection classification.dictionary_based._tde.histogram_intersection @MatthewMiddlehurst

Kernels

Name Class Maintainer References

Time series classifiers

Atoms

Univariate time series classifiers

Name Class Maintainer References
BOSS Ensemble classification.dictionary_based._boss.BOSSEnsemble @MatthewMiddlehurst Schäfer (2014) - The BOSS is concerned with time series classification in the presence of noise
BOSS Atom classification.dictionary_based._boss.IndividualBOSS @MatthewMiddlehurst
cBOSS classification.dictionary_based._cboss.ContractableBOSS @MatthewMiddlehurst Middlehurst et al (2019) - Scalable dictionary classifiers for time series classification
Temporal Dictionary Ensemble (TDE) classification.dictionary_based._tde.TemporalDictionaryEnsemble @MatthewMiddlehurst Middlehurst et al (2020) - The Temporal Dictionary Ensemble (TDE) Classifier for Time Series Classification
TDE Atom classification.dictionary_based._tde.IndividualTDE @MatthewMiddlehurst
Elastic Ensemble classification.distance_based._elastic_ensemble.ElasticEnsemble @jasonlines Lines, Bagnall (2015) - Time Series Classification with Ensembles of Elastic Distance Measures
Proximity Forest classification.distance_based._proximity_forest.ProximityForest @goastler Lucas et al (2019) - Proximity Forest: an effective and scalable distance-based classifier for time series
Proximity Stump classification.distance_based._proximity_forest.ProximityStump @goastler Lucas et al (2019) - Proximity Forest: an effective and scalable distance-based classifier for time series
Random Interval Spectral Forest (RISE) classification.frequency_based._rise.RandomIntervalSpectralForest @TonyBagnall Lines et al (2018) - Time Series Classification with HIVE-COTE: The Hierarchical Vote Collective of Transformation-Based Ensembles
Shapelet Transform Classifier classification.shapelet_based._stc.ShapeletTransformClassifier @TonyBagnall Hills et al (2014) - Classification of time series by shapelet transformation
Time Series Forest classification.interval_based._tsf.TimeSeriesForestClassifier @TonyBagnall Deng et al (2013) - A Time Series Forest for Classification and Feature Extraction
Time Series k-NN classification.distance_based._time_series_neighbors.KNeighborsTimeSeriesClassifier @jasonlines
Mr-SEQL classification.shapelet_based.mrseql.mrseql.MrSEQLClassifier @lnthach Interpretable Time Series Classification Using Linear Models and Multi-resolution Multi-domain Symbolic Representations
ShapeDTW classification.distance_based._shape_dtw.ShapeDTW @Multivin12 shapeDTW: Shape Dynamic Time Warping
WEASEL classification.dictionary_based._weasel.WEASEL @patrickZIB Fast and Accurate Time Series Classification with WEASEL
catch22 Forest Classifier contrib.hybrid._catch22_forest_classifier.Catch22ForestClassifier @MatthewMiddlehurst Lubba et al (2019) - catch22: CAnonical Time-series CHaracteristics
Canonical Interval Forest (CIF) contrib.interval_based.cif._CanonicalIntervalForest @MatthewMiddlehurst Middlehurst et al (2020) - The Canonical Interval Forest (CIF) Classifier for Time Series Classification

Multivariate time series classifiers

name sktime class maintainer literature
WEASEL+MUSE classifiers.dictionary_based.weasel.MUSE @patrickZIB Multivariate time series classification with WEASEL+ MUSE

Composition

Ensembling (abstract/1st order)

(only abstract ensembles in this list - hard-coded ensembles go in one of the lists for atoms)

Of Name Class Maintainer References
univariate TSC boosting TSC classifiers.compose.ensemble.TimeSeriesForestClassifier @mloning

Pipelines

Components Name Class Maintainer References
Transformers, classifiers, regressors pipeline sktime.pipeline.Pipeline

Reduction

From/output To/input Name Class Maintainer References
multivariate TSC univariate TSC column ensembler classifiers.compose.column_ensembler.ColumnEnsembleClassifier @abostrom

Time series regressors

Atoms

Univariate time series regressors

Name Class Maintainer References

Multivariate time series regressors

Name Class Maintainer References

Forecasting

Atoms

Endogenous time series forecasters

Name Class Maintainer References
Naive forecaster NaiveForecaster @mloning
Holt-Winters exponential smoothing forecaster ExpSmoothingForecaster @mloning, @big-o
Theta forecaster ThetaForecaster @big-o Unmasking the Theta method

Multivariate Time Series Forecasting

Name Class Maintainer References

Composition

Name Class Maintainer References

Pipeline

Name Class Maintainer References

Ensembling

Name Class Maintainer References
Online Hedge Ensemble Forecasting sktime.forecasting.online_ensemble.OnlineEnsembleForecaster @magittan A Parameter-free Hedging Algorithm

Forecasters

name sktime class maintainer literature

Reduction

Name Class Maintainer References