This presentation is to show how the underlying structure of a data set can impact the performance of an unsupervised learning algorithm. For this example, we will use the K-mean learning algorithm, and two simulated data sets from the Fundamental Clustering Problem Suite

In this data set, the K-mean algorithms(k=7) works very well, with a Rand Index= 1

However, with this data set the K-mean algorithms (k=2), does not works that well, with a Rand Index= 0.546