Clustergram Bug? - Different dendrograms produced by Clustergram vs SPSS and pdist -> linkage -> dendrogram using the same parameters

Good Afternoon,

I have tried already all different 'data processing' parameters of clustergram in order to get the same dendrogram that is obtained with pdist -> linkage -> dendrogram without any luck.

I also double-checked the results obtained by SPSS with the same data and parameteres used by pdist -> linkage -> dendrogram and the dendrograms are consistent.

Therefore I conclude is definitely something related with the clustergram class.

Could it be possible that there is a bug or something like that with that class?

To whom I should direct this question in order that it'll be further evaluated by the developers of clustergram?

Thanks in advance,

Diego

Answer 1

[Oleg Komarov]

You can contact technical support by compiling this form: http://www.mathworks.co.uk/support/contact_us/index.html

[Diego]

Thank you Oleg. I'll try the latest version first. It seems strange to me that there are not other complains about this, so chances are that the class is ok and it's just me. Thanks again, Regards, Diego

Answer 2

[Feiyu Xiong]

I think I have the same issue. I use pdist-linkage-dendrogram, then I use clustergram, the results are different.

[Feiyu Xiong]

I believe this is a bug and in 2013a they fixed this bug. I confirmed this by running my code on 2013a, and two function give the same cluster.