Highly comparable time-series analysis in Nitime

Dataset type: Imaging, Neuroscience
Data released on October 31, 2016

Fulcher BD (2016): Highly comparable time-series analysis in Nitime GigaScience Database. http://dx.doi.org/10.5524/100225

DOI10.5524/100225

The aim of this project was to demonstrate that an existing Matlab-based package for implementing thousands of time-series analysis methods, hctsa could be extended to a python-based implementation, for potential future inclusion into Nitime.
Recent work has contributed a comprehensive library of over 35,000 pieces of diverse time-series data, and over 7,000 unique structural features extracted from hundreds of different time-series analysis methods which can be explored through an associated website and implemented using the Matlab-based code package, hctsa.
The hctsa software provides a systematic, algorithmic platform for computing a wide range of structural properties from a single time series, including basic statistics of the distribution, linear correlation structure, stationarity, information theoretic and entropy measures, methods from the physical nonlinear time-series analysis literature, linear and nonlinear model fits, and others.
Thus, hctsa can be used to map a time series to a comprehensive vector of interpretable structural features and these features can then be systematically compared to determine and understand the most useful features for a given scientific objective (e.g., features of an EEG signal that help classify different patient groups).
In order to apply highly comparative time-series analysis in the neuroscience community, it would be desirable to implement some time-series analysis methods into Nitime, a python-based software package for performing time-series analysis on neuroscience data.
Implementation of useful time-series features into python, and potential integration with Nitime, would not only facilitate their use by the neuroscience community, but also their maintenance and development within an open source framework.
Although there are no plans to reimplement the full hctsa feature library in python, our hope is that published work describing useful time-series features (discovered using the hctsa library) can also contribute a python implementation, to promote its use by the neuroscience community.

Additional details

Read the peer-reviewed publication(s):


Related datasets:

doi:10.5524/100225 IsPartOf doi:10.5524/100215

Additional information:

https://github.com/benfulcher/hctsa_python





Displaying 1-1 of 1 File(s).
Date Action
October 31, 2016 Dataset publish
October 31, 2016 Description updated from : The aim of this project was to demonstrate that an existing Matlab-based package for implementing thousands of time-series analysis methods, hctsa could be extended to a python-based implementation, for potential future inclusion into Nitime.
Recent work has contributed a comprehensive library of over 35,000 pieces of diverse time-series data, and over 7,000 unique structural features extracted from hundreds of different time-series analysis methods which can be explored through an associated website and implemented using the Matlab-based code package, hctsa.
The hctsa software provides a systematic, algorithmic platform for computing a wide range of structural properties from a single time series, including basic statistics of the distribution, linear correlation structure, stationarity, information theoretic and entropy measures, methods from the physical nonlinear time-series analysis literature, linear and nonlinear model fits, and others.
Thus, hctsa can be used to map a time series to a comprehensive vector of interpretable structural features and these features can then be systematically compared to determine and understand the most useful features for a given scientific objective (e.g., features of an EEG signal that help classify different patient groups).
In order to apply highly comparative time-series analysis in the neuroscience community, it would be desirable to implement some time-series analysis methods into Nitime, a python-based software package for performing time-series analysis on neuroscience data.
Implementation of useful time-series features into python, and potential integration with Nitime, would not only facilitate their use by the neuroscience community, but also their maintenance and development within an open source framework.
Although there are no plans to reimplement the full hctsa feature library in python, our hope is that published work describing useful time-series features (discovered using the hctsa library) can also contribute a python implementation, to promote its use by the neuroscience community.
November 17, 2016 Manuscript Link added : 10.1186/s13742-016-0147-0