Programming the statistical procedures from SAS

Cluster analysis

Occasional Contributor
Posts: 8

Cluster analysis


I have 2 questions in my mind, can anyone help me on it.

1. Is it necessary to create dummy variables in cluster analysis? if yes
2. Do we need to standardize dummy variables along with other continuous variables to get ride of different measuring units

Trusted Advisor
Posts: 2,116

Re: Cluster analysis

The answers are in the PROC CLUSTER reference documentation.

1) cluster works only against numeric variables. So, if you have classification variables, you need to recode to dummies or apply some other sort of metric.
2) no, you don't need to standardize. See the examples.
Occasional Contributor
Posts: 8

Re: Cluster analysis

Thank you :-), now i have created dummy variables for the class variables, i will use them in clustering without standardizing.

however i have some outliers in 2 variables (1 continuous variable and 1 discrete variable), so do i need to do the outlier treatment first or stadadrization first.
Ask a Question
Discussion stats
  • 2 replies
  • 2 in conversation