Segmented Time Series Pipeline

This module is an sklearn compatible pipeline for machine learning time series data and sequences using a sliding window segmentation

class seglearn.pipe.Pype(steps, scorer=None, memory=None)[source]

This pipeline extends the sklearn Pipeline to support transformers that change X, y, sample_weight, and the number of samples.

It also adds some new options for setting hyper-parameters with callables and in reference to other parameters (see examples).

Parameters
stepslist

List of (name, transform) tuples (implementing fit/transform) that are chained, in the order in which they are chained, with the last object an estimator.

scorersklearn scorer object
memorycurrently not implemented

Examples

>>> from seglearn.transform import FeatureRep, SegmentX
>>> from seglearn.pipe import Pype
>>> from seglearn.datasets import load_watch
>>> from sklearn.ensemble import RandomForestClassifier
>>> from sklearn.preprocessing import StandardScaler
>>> data = load_watch()
>>> X = data['X']
>>> y = data['y']
>>> pipe = Pype([('segment', SegmentX()),
>>>              ('features', FeatureRep()),
>>>              ('scaler', StandardScaler()),
>>>              ('rf', RandomForestClassifier())])
>>> pipe.fit(X, y)
>>> print(pipe.score(X, y))
Attributes
N_trainnumber of training samples - available after calling fit method
N_testnumber of testing samples - available after calling predict, or score methods

Methods

decision_function(self, X)

Apply transforms, and decision_function of the final estimator

fit(self, X[, y])

Fit the model

fit_predict(self, X[, y])

Applies fit_predict of last step in pipeline after transforms.

fit_transform(self, X[, y])

Fit the model and transform with the final estimator Fits all the transforms one after the other and transforms the data, then uses fit_transform on transformed data with the final estimator.

get_params(self[, deep])

Get parameters for this estimator.

predict(self, X)

Apply transforms to the data, and predict with the final estimator

predict_as_series(self, X)

Returns predictions in a list, grouping predictions based on the series they were derived from

predict_log_proba(self, X)

Apply transforms, and predict_log_proba of the final estimator

predict_proba(self, X)

Apply transforms, and predict_proba of the final estimator

predict_unsegmented(self, X[, …])

Generates predictions for each time series on the same sampling as the original series, by resampling a prediction performed with sliding window segmentation

score(self, X[, y, sample_weight])

Apply transforms, and score with the final estimator

score_samples(self, X)

Apply transforms, and score_samples of the final estimator.

set_params(self, \*\*params)

Set the parameters of this estimator.

transform(self, X[, y])

Apply transforms, and transform with the final estimator This also works where final estimator is None: all prior transformations are applied.

transform_predict(self, X, y)

Apply transforms to the data, and predict with the final estimator.

decision_function(self, X)[source]

Apply transforms, and decision_function of the final estimator

Parameters
Xiterable

Data to predict on. Must fulfill input requirements of first step of the pipeline.

Returns
y_scorearray-like, shape = [n_samples, n_classes]
fit(self, X, y=None, **fit_params)[source]

Fit the model

Fit all the transforms one after the other and transform the data, then fit the transformed data using the final estimator.

Parameters
Xiterable

Training data. Must fulfill input requirements of first step of the pipeline.

yiterable, default=None

Training targets. Must fulfill label requirements for all steps of the pipeline.

**fit_paramsdict of string -> object

Parameters passed to the fit method of each step, where each parameter name is prefixed such that parameter p for step s has key s__p.

Returns
selfPipeline

This estimator

fit_transform(self, X, y=None, **fit_params)[source]

Fit the model and transform with the final estimator Fits all the transforms one after the other and transforms the data, then uses fit_transform on transformed data with the final estimator.

Parameters
Xiterable

Training data. Must fulfill input requirements of first step of the pipeline.

yiterable, default=None

Training targets. Must fulfill label requirements for all steps of the pipeline.

**fit_paramsdict of string -> object

Parameters passed to the fit method of each step, where each parameter name is prefixed such that parameter p for step s has key s__p.

Returns
Xtarray-like, shape = [n_samples, n_transformed_features]

Transformed samples

ytarray-like, shape = [n_samples]

Transformed target

predict(self, X)[source]

Apply transforms to the data, and predict with the final estimator

Parameters
Xiterable

Data to predict on. Must fulfill input requirements of first step of the pipeline.

Returns
yparray-like

Predicted transformed target

predict_as_series(self, X)[source]

Returns predictions in a list, grouping predictions based on the series they were derived from

Parameters
Xiterable

Data to predict on. Must fulfill input requirements of first step of the pipeline.

Returns
yplist

Predictions

predict_log_proba(self, X)[source]

Apply transforms, and predict_log_proba of the final estimator

Parameters
Xiterable

Data to predict on. Must fulfill input requirements of first step of the pipeline.

Returns
y_scorearray-like, shape = [n_samples, n_classes]
predict_proba(self, X)[source]

Apply transforms, and predict_proba of the final estimator

Parameters
Xiterable

Data to predict on. Must fulfill input requirements of first step of the pipeline.

Returns
y_probaarray-like, shape = [n_samples, n_classes]

Predicted probability of each class

predict_unsegmented(self, X, categorical_target=False)[source]

Generates predictions for each time series on the same sampling as the original series, by resampling a prediction performed with sliding window segmentation

Requires that one of the Segment transforms be part of the pipeline

See plot_feature_rep.py example

Parameters
Xiterable

Data to predict on. Must fulfill input requirements of first step of the pipeline.

categorical_targetboolean

Set to True for classification problems, and false for regression problems

Returns
ypiterable

Time series predictions on the same sampling as X

score(self, X, y=None, sample_weight=None)[source]

Apply transforms, and score with the final estimator

Parameters
Xiterable

Data to predict on. Must fulfill input requirements of first step of the pipeline.

yiterable, default=None

Targets used for scoring. Must fulfill label requirements for all steps of the pipeline.

sample_weightarray-like, default=None

If not None, this argument is passed as sample_weight keyword argument to the score method of the final estimator.

Returns
scorefloat
set_params(self, **params)[source]

Set the parameters of this estimator. Valid parameter keys can be listed with get_params().

Returns
self
transform(self, X, y=None)[source]

Apply transforms, and transform with the final estimator This also works where final estimator is None: all prior transformations are applied.

Parameters
Xiterable

Data to transform. Must fulfill input requirements of first step of the pipeline.

yarray-like

Target

Returns
Xtarray-like, shape = [n_samples, n_transformed_features]

Transformed data

ytarray-like, shape = [n_samples]

Transformed target

transform_predict(self, X, y)[source]

Apply transforms to the data, and predict with the final estimator. Unlike predict, this also returns the transformed target

Parameters
Xiterable

Data to predict on. Must fulfill input requirements of first step of the pipeline.

yarray-like

target

Returns
ytarray-like

Transformed target

yparray-like

Predicted transformed target