# Approximating time series¶

Approximation algorithms try to capture the most important information from time series. They can be seen as simple feature extraction algorithms.

Piecewise Aggregate Approximation

Symbolic Aggregate approXimation

# Bag-of-words transformation¶

Bag-of-words algorithms transform a sequence of symbols into a bag of words.

# Classification algorithms¶

Classification algorithms can directly classify raw time series.

Bag-of-SFA Symbols in Vector Space (BOSSVS)

Learning Time-Series Shapelets

Symbolic Aggregate approXimation in Vector Space Model (SAX-VSM)

# Clustering time series¶

Depending on a suitable metric, it is possible to cluster time series.

Time Series Clustering with DTW and BOSS

# Dataset utilities¶

Examples on how to load and make time series datasets.

Making a Cylinder-Bell-Funnel dataset

# Decomposing time series¶

Decomposition algorithms decompose time series into several components.

Trend-Seasonal decomposition with Singular Spectrum Analysis

# Imaging time series¶

Imaging algorithms transform time series into images.

Data set of Gramian angular fields

Data set of Markov transition fields

Single Markov transition field

# Metrics¶

Specific metrics for time series have been developed. The examples below illustrate some of the implemented metrics.

# Multivariate time series¶

Specific algorithms for multivariate time series have been developed. The examples below illustrate some of the implemented ones.

# Preprocessing tools¶

Preprocessing data is a common task in machine learning. The examples below illustrate the preprocessing tools available in this module.

# Transformation algorithms¶

Transformation algorithms try to capture the most important information from time series using advanced transformation. They can be seen as complex feature extraction algorithms.

RandOm Convolutional KErnel Transform (ROCKET)

Word ExtrAction for time SEries cLassification (WEASEL)