Unsupervised analysis is looking for previously undetected patterns in a
data, usually those, we are not aware of. Our method is grouping the
data into a clusters with k-Means method. This approach helps us to
identify commonalities in the data, and finally helps us detect
anomalous data points that do not fit into previous identified clusters.
The optimal number of clusters is calculated as the intersection of two linear curves: the first line is a linear regression of initial gain logarithmic values, the second line is a linear regression of last logarithmic gain values.
