classification nominal/ordinal data
dan76
New Altair Community Member
Hello,
i'm still very new on RM and need some advice ::).
i have data from a survey. about 50.000cases and 30 attributes. my target is to look for pattern in this data. I think i have to use unsupervised techniques such as clustering.
As i know, kmeans needs numerical attributes for clustering, but my data consists of nominal and ordinal and there is no distance.
Thanks for help and sry abot this worse english.
Dan
i'm still very new on RM and need some advice ::).
i have data from a survey. about 50.000cases and 30 attributes. my target is to look for pattern in this data. I think i have to use unsupervised techniques such as clustering.
As i know, kmeans needs numerical attributes for clustering, but my data consists of nominal and ordinal and there is no distance.
Thanks for help and sry abot this worse english.
Dan
Tagged:
0
Answers
-
Hello
As a short suggestion:
Try the operators KMedoids or AgglomerativeClustering. These provide more distance measures to choose from.
regards,
Steffen
0