clarification regarding "silhoutte" plot

I have extracted GLCM, HOG, wavelet, Power spectral and Gabor Features for three classes. I want to do correlation analysis for these features. for correlation analysis I have plotted silhoutte coefficient for all the above features. but I'm not getting how to analyze these plots, which means which feature is correlated good (for which feature silhoutte plot is good) and I need brief detail about the plots (what are those neagative and positive side of the plots).

the three classes are metal pipe, steel box and plastic box. purpose is buried object detection and classification.

I have attached the silhoutte plots for all the above mentioned features.

[santhosh kumar buddepu]

please answer this question from matlab community it will help me alot

Answer 1

[the cyclist]

The Wikipedia page for silhouette clustering has a pretty good explanation, if you read it carefully and understand it.

For each cluster, a value closer to 1 means that that point is well matched to its cluser. A value closer to -1 means that it is poorly matched to its cluster.

The MATLAB documentation for the silhouette function shows an example of what good clustering looks like. (Notice that it has no negative values at all.)