k-Means Clustering in EEG (Brain Waves) Timeseries

Clustering is a process of assigning data points into groups (clusters). The goal is to assign “similar” points to the same cluster, while “different” points belong to distinct clusters. k-means and its modification form a very important and commonly used group of clustering methods. k-means is fast, simple and can be easily implemented in any programming package. Classical k-means leads to a local minimum of a function that represents the sum of the squares of the Euclidean distances between all the points and the centres of the clusters where the points were assigned. Therefore, the mathematical interpretation of k-means is straigthforward. In this paper, we use k-means methods and their modifications to produce clusters in timeseries data and use these clusters to create classification rules for distinguishing between the classes. The results of numerical experiments demonstrate that these fast and simple approaches provide high classification accuracy.