Pattern Transition Detection Algorithm (PTDA)

This algorithm is an extension of the change point analysis to detect general changes in the pattern of a time series.

Version 1.1.3 (997 ko)
397 téléchargements
(2)
3 sept. 2021

Vous suivez désormais cette soumission

The algorithm calculates the Time Frequency Distribution (TFD), Recurrence Plot (RP), and Dynamic Complexity (DC) of a time series and applies the change point analysis to the original as well as to the time series/matrices of TFD, RP and DC.
Details can be found in our accompanying paper (submitted for publication). The idea of the algorithm is presented in this article: https://www.frontiersin.org/articles/10.3389/fpsyg.2020.01970/full.
**************************************************************************
Updates Version 1.1
- outliers are removed from the input time series
- function "findchangepts" is used instead of "ischange"
- the original time series is also assessed for changes of a linear trend
- recurrence plots, dynamic complexity, and time frequency distributions are not assessed for changes of the variance anymore (only changes of the mean)
- omitted outlier deletion of the resulting change points
- cross-validation omitted (not necessary anymore)
- peakfinder instead of mean used to determine overall transition point
- added visualization for the results
- added example data

Citation pour cette source

Kathrin Viol (2026). Pattern Transition Detection Algorithm (PTDA) (https://fr.mathworks.com/matlabcentral/fileexchange/80380-pattern-transition-detection-algorithm-ptda), MATLAB Central File Exchange. Extrait(e) le .

Informations générales

Compatibilité avec les versions de MATLAB

  • Compatible avec R2018b

Plateformes compatibles

  • Windows
  • macOS
  • Linux
Ouvrir dans un nouvel onglet
Version Publié le Notes de version
1.1.3

Bugfix of the entry and statistics of the overall transition point if more than one time series is entered.

1.1.2

Bug fixed for calculating the statistics.
1.1.1

corrected file name
1.1

- several methodological refinements (see Description)
- multiple variables of the same system can be assessed together, resulting in an overall transition point for the whole system
- added visualization for the results
- added example dataset
1.0.2

Output text changed from "change point" to "transition point".

1.0.1

Minor fixing: The algorithm produced an error message and aborted when a time series consists of 1 or 2 values only. This was fixed by omitting the calculation of the Dynamic Complexity in those cases, accompanied by a corresponding message.
1.0.0