K-Shape clustering^•

Last modified on 01 Oct 2021.

What?

K-Shape clustering method is a method for clustering time series.

The main ideas of this algorithms are:

• The distance measure is based on the cross-correlation of two time series.
• The clustering process uses iterative approaches.
• It is a raw-based method. It is fundamentally a variant of k-means with some interesting modification.
  • Define a distance between 2 time series.
  • How to average multiple time series.

References

• (article) k-Shape: Efficient and Accurate Clustering of Time Series
• Ryan’s blog – K-Shape Clustering Algorithm

×Close