Error when running kolter's algorithm

beckel / nilm-eval

NILM-EVAL: An evaluation framework for non-intrusive load monitoring algorithms

GNU General Public License v2.0

111 stars 50 forks source link

Error when running kolter's algorithm #7

Open yesir1006 opened 7 years ago

yesir1006 commented 7 years ago

Hi,

I run kolter's algorithm with default configuration and experiment, hover, I got an error:

Error using kmeans (line 159) KMEANS does not accept complex data.

Error in SpectralClustering (line 66) C = kmeans(U, k, 'start', 'sample', ...

Error in kolter (line 103) clusters = SpectralClustering(adjacencyMatrix, numOfClusters, 1); ...

I use matlab2016a and wonder if it is the version problem. Did any one meet this problem? Any help would be appreciated. Thank you.

HajerSalem commented 6 years ago

Hi Yesir, Did you figure out why it is happening. I am getting the same issue. Thank you,

yesir1006 commented 6 years ago

Hi HajerSalem, Unfortunately not yet. I am working on other subject recently, the problem is still remained.

marinadorokhova commented 6 years ago

I have now exactly the same error, didn't resolve it yet.

habring commented 1 year ago

I have the same issue: Problem is the following:

The error appears in the function kmeans used in SpectralClustering because the first input argument of kmeans is not real. This first input argument are the eigenvectors of the Laplacian L= D-W. Apparently in the setting at hand the adjacency matrix W is not symmetric and, thus, in general D might have complex eigenvectors. If the adjacency matrix is allowed to be non-symmetric, then I think this whole method might have a flaw. Otherwise I assume there is a bug in the computation of the adjacency matrix.