How can I do a hierarchical clustering with correlation as distance?

4 次查看（过去 30 天）

Michael 2015-10-23

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/250648-how-can-i-do-a-hierarchical-clustering-with-correlation-as-distance

I have a set of monthly financial returns data for a few stocks. I have also computed the pairwise correlation between all these. I am told to do a hierarchical cluster analysis to assign these stocks to groups so as to minimize the average within group distance between group members. Distance is to be measured as 1 minus the correlation coefficient between the two stock returns.

I have found the matlab function clusterdata which also supports 'distance' with the option 'correlation' that explains: One minus the sample correlation between points (treated as sequences of values)

I cannot understand how I can pass my data to this function and that 'sequence of values' to be passed. Should I pass the correlations or the raw data? And how am I to extract groupings for each stock?

Thank you