so i have a correlation matrix, and based on the correlation values i would like to form various clusters, lets say n. from those n clusters, 1 member (attribute) would be selected and those n attributes would be used to form a model in combination with the other attributes with low correlations, and then we would try different combinations from the clusters.
for eg:-
total attributes = 20
cluster 1- attr4, attr6, attr7
cluster 2- attr9, attr14, attr16
therefore the 1st model would be trained on attr4 (from cluster 1) and attr9 (from cluster 2) and rest of other attributes which arent in the cluster.
so the 2nd model would be trained on attr4 (from cluster 1) and attr14 (from cluster 2) and rest of other attributes which arent in the cluster.
i cant figure out how to find and store those clusters from the correlation matrix. any kind of help is appreciated.
