How to use similarity matrix in hclust

Asked Jul 23 '15 at 14:54

Active Jul 23 '15 at 23:12

Viewed 645 times

I have a matrix which represents similarity between nodes, how can I use it in hclust method which takes dissimilarity matrices ?

does setting dissimilarity to +1 and similarity to negative values solves the problem ? for example...

 bank.albert    bank.gloucester bank.laurier    bank.lisgar
bank-albert 0   1   1   1
bank-gloucester 1   0   -5036.625   1
bank-laurier    1   -5036.625   0   1
bank-lisgar 1   1   1   0
bank-nepean 1   -6524.875   1   -7438.375

edited Jul 23 '15 at 23:12

asked Jul 23 '15 at 14:54

Exorcismus

2,243
1
35
68

1

I your similarity is scaled to [0,1] (and symmetric), you could use `dissimilarity = 1 - similarity`. – EDi Jul 23 '15 at 16:00
@EDi please check the edit – Exorcismus Jul 23 '15 at 23:13

How to use similarity matrix in hclust

0 Answers0