Transformations

The aeon.transformations module contains classes for series transformations. The module is organised into CollectionTransformers which transform a collection of time series into a different representation and SeriesTransformers which transform single time series.

All transformers in aeon can be listed using the aeon.utils.discovery .all_estimators function using type_filter="transformer", optionally filtered by tags. Valid tags for transformers can be found with aeon.utils.tags. all_tags_for_estimator function with the argument "transformer".

Collection transformers

AutocorrelationFunctionTransformer([n_lags, ...])	Autocorrelation function transformer.
ARCoefficientTransformer([order, ...])	Autoreggression coefficient feature transformer.
Centerer()	Centering transformer for collections.
DownsampleTransformer([downsample_by, ...])	Downsample the time dimension of a collection of time series.
DWTTransformer([n_levels])	Discrete Wavelet Transform Transformer.
HOG1DTransformer([n_intervals, n_bins, ...])	HOG1D transform.
MatrixProfile(*args, **kwargs)	Return the matrix profile and index profile for each time series of a dataset.
MinMaxScaler([min, max])	MinMax transformer for collections.
Normalizer()	Normaliser transformer for collections.
PeriodogramTransformer([pad_series, ...])	Periodogram transformer.
SeriesToCollectionBroadcaster(transformer)	Broadcast a BaseSeriesTransformer over a collection of time series.
SlopeTransformer([n_intervals])	Piecewise slope transformation.
SimpleImputer([strategy, fill_value])	Time series imputer.
Tabularizer()	A transformer that turns time series collection into tabular data.

Channel selection

ChannelScorer(estimator[, scoring_function, ...])	Performs channel selection using a single channel classifier or regressor.
ElbowClassPairwise([distance, ...])	Elbow Class Pairwise (ECP) transformer to select a subset of channels.
ElbowClassSum([distance, prototype_type, ...])	Elbow Class Sum (ECS) transformer to select a subset of channels/variables.
RandomChannelSelector([p, random_state])	Selects a random proportion of channels.

Compose

CollectionTransformerPipeline(transformers)	Pipeline of collection transformers.
CollectionId()	Identity transformer, returns data unchanged in transform/inverse_transform.

Convolution based

Rocket([n_kernels, normalise, n_jobs, ...])	RandOm Convolutional KErnel Transform (ROCKET).
MiniRocket([n_kernels, ...])	MINImally RandOm Convolutional KErnel Transform (MiniRocket).
MultiRocket([n_kernels, ...])	Multi RandOm Convolutional KErnel Transform (MultiRocket).
HydraTransformer([n_kernels, n_groups, ...])	Hydra Transformer.

ROCKETGPU([n_kernels, kernel_size, padding, ...])

RandOm Convolutional KErnel Transform (ROCKET) for GPU.

Dictionary-based features

SAX([n_segments, alphabet_size, ...])	Symbolic Aggregate approXimation (SAX) transformer.
PAA([n_segments])	Piecewise Aggregate Approximation Transformer (PAA).
SFA([word_length, alphabet_size, ...])	Symbolic Fourier Approximation (SFA) Transformer.
SFAFast([word_length, alphabet_size, ...])	Symbolic Fourier Approximation (SFA) Transformer.
SFAWhole([word_length, alphabet_size, norm, ...])	Symbolic Fourier Approximation (SFA) Transformer.
BORF([window_size_min_window_size, ...])	Bag-of-Receptive-Fields (BORF) Transformer.

Feature based

Catch22([features, catch24, outlier_norm, ...])	Canonical Time-series Characteristics (Catch22).
TSFresh([default_fc_parameters, ...])	Transformer for extracting time series features via tsfresh.extract_features.
TSFreshRelevant([default_fc_parameters, ...])	Transformer for extracting time series features via tsfresh.extract_features.
SevenNumberSummary([summary_stats])	Seven-number summary transformer.

Imbalance

ADASYN([n_neighbors, random_state, ...])	Adaptive Synthetic Sampling (ADASYN) over-sampler.
ESMOTE([n_neighbors, distance, ...])	Elastic Synthetic Minority Over-sampling Technique (ESMOTE).
SMOTE([n_neighbors, random_state, distance, ...])	Synthetic Minority Over-sampling TEchnique (SMOTE) for imbalanced datasets.
OHIT([k, kapa, drT, distance, random_state])	Over-sampling based on High-density region and Iterative Thresholding (OHIT).

Interval based

RandomIntervals([n_intervals, ...])	Random interval feature transformer.
SupervisedIntervals([n_intervals, ...])	Supervised interval feature transformer.
QUANTTransformer([interval_depth, ...])	QUANT interval transform.

Self Supervised

TRILITE([alpha, weight_ref_min, ...])	TRIplet Loss In TimE (TRILITE).
TimeMCL([alpha, mixup_temperature, ...])	Time Mixup Contrastive Learning (TimeMCL).

Shapelet based

RandomShapeletTransform([...])	Random Shapelet Transform.
RandomDilatedShapeletTransform([...])	Random Dilated Shapelet Transform (RDST) as described in [R1a26faa97573-1], [R1a26faa97573-2].
SAST([lengths, stride, nb_inst_per_class, ...])	Scalable and Accurate Subsequence Transform (SAST).
RSAST([n_random_points, len_method, ...])	Random Scalable and Accurate Subsequence Transform (RSAST).

Signature based

SignatureTransformer([augmentation_list, ...])

Transformation class from the signature method.

Unequal length

Padder([padded_length, fill_value, ...])	Pad unequal length time series to equal, fixed length.
Resizer([resized_length])	Resize unequal length time series to equal, fixed length.
Truncator([truncated_length, error_on_short])	Truncate unequal length time series to equal, fixed length.

Series transforms

AutoCorrelationSeriesTransformer([n_lags])	Auto-correlation transformer.
ClaSPTransformer([window_length, ...])	ClaSP (Classification Score Profile) Transformer.
DifferenceTransformer([order])	Calculates the n-th order difference of a time series.
Dobin([frac, k])	Distance based Outlier BasIs using Neighbors (DOBIN).
MatrixProfileSeriesTransformer(*args, **kwargs)	Calculate the matrix profile of a time series.
MatrixProfileTransformer([window_length])	Calculate the matrix profile of a time series.
LogTransformer([offset, scale])	Natural logarithm transformation.
PLASeriesTransformer([max_error, ...])	Piecewise Linear Approximation (PLA) for time series transformation.
StatsModelsACF([adjusted, n_lags, fft, missing])	Auto-correlation wrapper for statsmodels.
StatsModelsPACF([n_lags, method])	Partial auto-correlation wrapper for statsmodels.
BKFilter([low, high, K])	Filter a times series using the Baxter-King filter.
BoxCoxTransformer([bounds, method, sp])	Box-Cox power transform.
ScaledLogitSeriesTransformer([lower_bound, ...])	Scaled logit transform or Log transform.
PCASeriesTransformer([n_components, copy, ...])	Principal Components Analysis applied as transformer.
WarpingSeriesTransformer([series_index, ...])	Warping Path Transformer.

Compose

SeriesTransformerPipeline(transformers)	Pipeline of series transformers.
SeriesId()	Identity transformer, returns data unchanged in transform/inverse_transform.

Smoothing

DiscreteFourierApproximation([r, sort])	Filter a times series using a Discrete Fourier Approximation.
ExponentialSmoothing([alpha, window_size])	Filter a time series using exponential smoothing.
GaussianFilter([sigma, order])	Filter a time series using Gaussian filter.
MovingAverage([window_size])	Calculate the moving average for a time series.
LOWESS([frac, it, delta])	LOWESS smoother for equally spaced time series [Rdb89580bbd2d-1].
SavitzkyGolayFilter([window_length, polyorder])	Filter a times series using Savitzky-Golay (SG).
RecursiveMedianSieve([window_length])	Filter a times series using a Recursive Median Sieve.

Base

BaseCollectionTransformer()

Transformer base class for collections.

BaseSeriesTransformer(axis)

Transformer base class for collections.

BaseTransformer()

Transformer base class.