search

aeon-toolkit / aeon

A toolkit for machine learning from time series
https://aeon-toolkit.org/
BSD 3-Clause "New" or "Revised" License
979 stars 118 forks source link

[ENH] Adaptive linear segmentation algorithm #948

Open TonyBagnall opened 10 months ago

TonyBagnall commented 10 months ago

Describe the feature or idea you want to propose

It would be great to have an implementation of this piecewise linear approximation method for segmenting time series

https://ieeexplore.ieee.org/abstract/document/989531

Describe your proposed solution

Implement as a base segmenter, internally store the fitted slopes

Describe alternatives you've considered, if relevant

No response

Additional context

No response

Moonzyyy commented 3 months ago

Will be taking doing this now!

patrickzib commented 3 months ago

This approach, in my assessment, does not fall under the category of segmentation algorithms such as "change point detection" or "semantic segmentation." Rather than being classified among segmentation techniques, it aligns more closely with data representation methods like Piecewise Aggregate Approximation (PAA), Symbolic Aggregate approXimation (SAX), or Symbolic Fourier Approximation (SFA). These methods focus on summarizing or approximating time series data rather than identifying distinct segments or change points within the data.