Library
How does the BayesiaLab K-Means discretization algorithm work?
Quote 0 0
Library
This discretization algorithm is an unsupervised univariate discretization algorithm that consists of applying the classical K-means clustering to one-dimensional continuous data.The Expectation-Maximization algorithm works as follows:[list=1gv2qs7n][*gv2qs7n]Initialization: random creation of K centers[/*:mgv2qs7n][*gv2qs7n]Expectation: each point is associated with the closest center[/*:mgv2qs7n][*gv2qs7n]Maximization: each center position is computed as the barycenter of its associated points[/*:mgv2qs7n][/listgv2qs7n]Steps 2 and 3 are repeated until convergence is reached.The discretization thresholds used by BayesiaLab are defined as:[latexgv2qs7n]T_i = \frac{K_{i}+K_{i+1}}{2}[/latexgv2qs7n]The figure below illustrates how this algorithm works with K=3.kMeans.png 
Quote 0 0