What to do in cluster analysis

What to do in cluster analysis

1. 1. Identify the subjects or objects (variables) to be clustered.

   • 2. Select the distance measure for computing distance between cases.
     • 3. Hierarchical clustering
       • 3.1. Decide the number of clusters to be formed.
       • 3.2. Map and interpret the clusters.
       • 4. K-M clustering
         • 4.1. Decide the number of clusters to be formed.
         • 4.2. Map and interpret the clusters.
         • 5. Asess reliability and validity of the obtained clusters
           • Apply the cluster analysis on the same data by using different distance measure.
           • Apply the cluster analysis on the same data by using different clustering technique.
           • Split the same data randomly into two halves and apply the cluster analysis separately on each part.
           • Repeat cluster analysis on same data several times by deleting one variable each time
           • Repeat cluster analysis several times, using a different order each time.