This code is a study into the Highly Comparative Time Series Analysis (HCTSA) library developed by Ben Fulcher - https://github.com/benfulcher/hctsa.
It provides a method of selecting a reduced set of operations from the full library available, using the operations' empirical behaviour on a widely varied data set.
- Ensure that you have fully installed the HCTSA library from the above link - and the external toolboxes required
- Run
startup.m
from the HCTSA library each time before you use this code to add the HCTSA functions to your path
Ensure your dataset is from as widely varied sources as possible to ensure the most generally useful operations are given. Save this dataset in the standard HCTSA format
To start the analysis run the code from SS_Clustering_main
- The separate functions can be used to reproduce individual steps separately
- However if changing an important parameter (e.g. K for the k-medoids method) it is recommended to re-run the whole
SS_Clustering_main
script to ensure all saved files are in sync.