RapidMiner AutoModel k-means summary

tonyboy9

I need to interpret the k-means summary. I have no idea what the jumble of facts mean under each segment. How do I locate which segment has the problem attribute(s) I need to see. If I can see the applicable segment, what does that mean in terms of problem solving? 

jacobcybulski

The summary tells you more or less the same as the centroid chart. The summary tells you the characteristics of each cluster by comparing the most important attributes (with highest variance) against the average of the same attribute in a data set. If you check the summary description against the centroid chart (each cluster shows the values of its centroids), you will see that when the summary states the attribute value is say 75% less than the average, in the centroid chart the cluster line will dip for that attribute as compared against other cluster lines. So in both cases you have a description of each cluster in terms of its attribute values.