seglearn 1.0.4

A template for scikit-learn compatible packages.

seglearn

Seglearn is a python package for machine learning time series or sequences. It provides an integrated pipeline for segmentation, feature extraction, feature processing, and final estimator. Seglearn provides a flexible approach to multivariate time series and related contextual (meta) data for classification, regression, and forecasting problems. Support and examples are provided for learning time series with classical machine learning and deep learning models. It is compatible with scikit-learn.

Documentation

Installation documentation, API documentation, and examples can be found on the documentation.

Dependencies

seglearn is tested to work under Python 2.7 and Python 3.5. The dependency requirements are based on the last scikit-learn release:

• scipy(>=0.13.3)

• numpy(>=1.8.2)

• scikit-learn(>=0.19.0)

Additionally, to run the examples, you need:

• matplotlib(>=2.0.0)

• keras (>=2.1.4) for the neural network examples

• pandas

In order to run the test cases, you need:

• pytest

The neural network examples were tested on keras using the tensorflow-gpu backend, which is recommended.

Installation

seglearn-learn is currently available on the PyPi’s repository and you can install it via pip:

pip install -U seglearn

or if you use python3:

pip3 install -U seglearn

If you prefer, you can clone it and run the setup.py file. Use the following commands to get a copy from GitHub and install all dependencies:

git clone https://github.com/dmbee/seglearn.git
cd seglearn
pip install .

Or install using pip and GitHub:

pip install -U git+https://github.com/dmbee/seglearn.git

Testing

After installation, you can use pytest to run the test suite from seglearn’s root directory:

pytest

Change Log

Version 1.0.4:

• Fixed bug with Interp class and improved its unit testing

• Added preprocessing module with TargetRunLengthEncoder transform

• Further features functions contributed by Matthias Gazzari (qtux)

Version 1.0.3:

• Continuous integration fixes

• Documentation and unit testing improvements

Version 1.0.2:

• Added further features

Version 1.0.1:

• Improvements to documentation, error checking, and unit tests

• feature_functions min and max changed to minimum and maximum respectively

Version 1.0.0:

• Major API change integrating preprocessing and estimation into single pipeline (Pype)

• Note this is unfortunately not backwards compatible

• Added Interp transformer for resampling time series

Version 0.2.0:

• Added transformer for segmenting data set with padding and / or truncation

• Implemented random_state initialization for SegPipe

• Corrected documentation for FeatureRep

Version 0.1.6:

• First stable version

Development

The development of this scikit-learn-contrib is in line with the one of the scikit-learn community. Therefore, you can refer to their Development Guide.

Please submit new pull requests on the dev branch with unit tests and an example to demonstrate any new functionality / api changes.

About

This package was developed by:

David M. Burns MD, PhD(c)
Sunnybrook Research Institute
University of Toronto
Email: [e-mail]

Citing seglearn

If you use seglearn in a scientific publication, we would appreciate citations to the following paper:

@article{arXiv:1803.08118
author  = {David Burns, Cari Whyne},
title   = {Seglearn: A Python Package for Learning Sequences and Time Series},
journal = {arXiv},
year    = {2018},
url     = {https://arxiv.org/abs/1803.08118}
}

If you use the seglearn test data in a scientific publication, we would appreciate citations to the following paper:

@article{arXiv:1802.01489
author  = {David Burns, Nathan Leung, Michael Hardisty, Cari Whyne, Patrick Henry, Stewart McLachlin},
title   = {Shoulder Physiotherapy Exercise Recognition: Machine Learning the Inertial Signals from a Smartwatch},
journal = {arXiv},
year    = {2018},
url     = {https://arxiv.org/abs/1802.01489}
}