Highly comparative time-series analysis using hcts
GitHub: hctsa
Search…
Introduction
List of included code files
Getting Started
Advice and common pitfalls
Installing and setting up
Running hctsa computations
Analyzing and visualizing results
Assigning group labels to data
Filtering and normalizing
Clustering rows and columns
Visualizing the data matrix
Plotting the time series
Low dimensional representation
Finding nearest neighbors
Investigating specific operations
Exploring classification accuracy
Finding informative features
Interpreting features
Comparing to existing features
Working with short time series
Working with a mySQL database
Powered By GitBook
Clustering rows and columns
For the purposes of visualizing the data matrix, it is often desirable to have the rows and columns reordered to put similar rows adjacent to one another, and similarly to place similar columns adjacent to one another. This reordering can be done using hierarchical linkage clustering, by the function TS_Cluster:
distanceMetricRow = 'euclidean'; % time-series feature distance
linkageMethodRow = 'average'; % linkage method
distanceMetricCol = 'corr_fast'; % a (poor) approximation of correlations with NaNs
linkageMethodCol = 'average'; % linkage method
​
TS_Cluster(distanceMetricRow, linkageMethodRow, distanceMetricCol, linkageMethodCol);
This function reads in the data from HCTSA_N.mat, and stores the re-ordering of rows and columns back into HCTSA_N.mat in the ts_clust and op_clust (and, if the size is manageable, also the pairwise distance information). Visualization functions (such as TS_PlotDataMatrix and TS_SimSearch) can then take advantage of this information, using the general input label 'cl'.
Previous
Filtering and normalizing
Next
Visualizing the data matrix
Last modified 2yr ago
Copy link