Skip to content

Latest commit

 

History

History

phasespace_svm

PhaseSpace-SVM

Citekey MaPerkins2003Timeseries
Source Code own
Learning type unsupervised
Input dimensionality univariate

Notes

  • We use left-aligned windows to embed the time series into a phase space, because it produces better results.

    Right-aligned embedding: right

    Left-aligned embedding: left

  • We take the raw scores from the SVM as output instead of aggregated binary labels. The scores are aggregated using sum().

Requirements

  • numpy
  • sklearn