Cluster Model Visualizer

User134999User134999 Member Posts: 1 Contributor I
I would like to know how to interpret the results from Cluster Model Visualizer.  If we consider the Tutorial Process from operator description, we obtain for Cluster_0 "a1 is on average 65.48% larger". 
The averege of a1 for all rows is 5.843333333; the average of a1 for Cluster_0 is 6.854.
Then what this percent (65.48%) means? How it is computied? 

Best Answer

  • Options
    jmerglerjmergler Administrator, Moderator, Employee, RapidMiner Certified Analyst, Member, University Professor Posts: 41 Guru
    edited January 2023 Solution Accepted
    Hi User134999,
    The calculation takes the value of the cluster's centroid at that attribute, the range of the attribute and the mean of the attribute across all clusters. It normalizes both the value of the centroid and the overall average, over the full attribute range. Then it computes the difference between the two divided by the overall normalized average. The interpretation is the scaled or percent difference between the cluster centroid and the overall mean for that attribute. The centroid is a part of the cluster model that was previously computed.
    There are some differences for rounding, but in this example it is roughly (((6.9 - 4.3)/3.6)-((5.8-4.3)/3.6))/((5.8-4.3)/3.6)
Sign In or Register to comment.