Overall Results: I

In nearly every one of these k=50 runs (180 of them, i.e., 60 targets with three different clustering methods) there were just three distinct clusters of coefficients

There was a high degree of commonality in the constitution of these three clusters
- Some coefficients always appeared together

Very similar results obtained in experiments with k=100, 200 and 400 and with the IDAlert database

In nearly every one of these k=50 runs (180 of them, i.e., 60 targets with three different clustering methods) there were just three distinct clusters of coefficients