Fig. 3
From: GMMchi: gene expression clustering using Gaussian mixture modeling
Dynamic binning: based on simulated data that creates a bimodal distribution. The mean and variance of the simulated 400 random samples in each histogram are 5, 10, and 3, 10, respectively. On the left is the histogram before applying dynamic binning. Samples within each bin in the far left and right of the left histogram are less than 5 thus falling below the prerequisite of applying the chi-square goodness of fit test. On the right, dynamic binning solves this by dynamically combining bins without over-manipulating the overall distribution. In this example, only 15 samples were combined via dynamic binning. The bins combined via dynamic binning are boxed in red dashes