| 12345678910111213141516171819202122232425262728293031323334 |
- function [vecCl,vecClI,nCl,sCl] = findClu(hV,tStat,lenMin)
- % IDENTIFY CLUSTERS IN OBSERVED DATA
- % This function identifies clusters in the observed data, based on the
- % given minimum length of consecutive points. It also counts the number
- % of clusters and returns the size of each identified cluster
- % create vector that contains all clusters
- % code from forum: replace ones by number of consecutive repetitions
- d = [1,diff(hV)~=0,1]; % TRUE if values change
- n = diff(find(d)); % Number of repetitions
- X = repelem(n,n);
- vecClI = X.*hV;
- vecClI(vecClI<lenMin) = 0; % replace numbers that are smaller than criterion (less than X consecutive repetitions) by 0
- vecClI(vecClI>=lenMin) = 1; % replace numbers that are equal or larger than criterion (X consecutive repetitions or more) by 1
- % index into t-value vector to extract clusters (based on
- % defined criterion)
- vecCl = tStat.*vecClI;
- % count number of clusters in vector
- vecClIs = [0,vecClI]; % add 0 at start to make sure possible cluster starting at point 1 is considered
- vecClIe = [vecClI,0]; % add 0 at the end to make sure possible cluster lasting until the end is considered
- startCl = find(diff(vecClIs)==1); % find all points where difference vector is 1 (starting points of clusters in original vector)
- endCl = find(diff(vecClIe)==-1); % find all points where difference vector is -1 (ending points of clusters in original vector)
- nCl = numel(startCl); % return number of cluster starting points = number of clusters
- % calculate size of each cluster
- sCl = zeros(nCl,1);
- for c = 1:nCl
- sCl(c) = sum(vecCl(startCl(c):endCl(c))); % calulate size of each identified cluster
- end
- end
|
