I am currently trying to interpret a set of results gleaned from running SimpleKMeans clustering on the Diabetes.arff data set.
https://i.sstatic.net/T4eho.jpg - link to clustered instances (figure 1)
So far I can understand that the clustered instances (figure 1) show that 500 variables have been classified as tested negative and 268 have been classified as tested positive.
https://i.sstatic.net/X9szt.jpg - link to groundtruth values (figure 2)
When these values are compared to the groundtruth values, there is not much difference because the correct clustering should have shown that 500 are to be classified as tested negative and 268 were to be classified as tested positive. This technically means that the SimpleKMeans clustering approach is suited for this data set as it has correctly classified the instances.
But I do not know how to interpret the information in the Cluster centroids table, under he Full Data, Cluster #0 and Cluster 1 headings. What do they tell us about the data set (figure 1)?
You should drop the class attribute before you do clustering. It has too much predictive power, and as a consequence of this, the clustering algorithm has a strong bias to prefer the class attribute internally.
You can do this attribute removal in the "Preprocess" panel by clicking the "remove" button, or in the "Cluster" panel by clicking "ignore attributes", then selecting "class" attribute.
Then cluster again. I suggest to start with k = 2, which is the number of unique values of the "class" attribute. (And then check if the cluster assignments correspond to the original attribute, Or do something else.)
By the way, seems to me that you were not working on the "glass" data set, but on the "diabetes" dataset.
