1
عمومی::
گوسی
Figure 1: Didactic example of n = 150 data points xj R2 sampled from three bivariate Gaussian distributions.
Gaussian data5.
As k-means is tailored to Gaussian mixtures and since the initial choice of centroids was favorable, it took only four iterations to converge to this rea- sonable solution.
As it implicitly as- sumes locally Gaussian data, the algo-
For our ex- ample, the transformation is too naïve to have a noticeable effect on the result; the data are simply not Gaussian enough for k-means to identify their "true" latent substructures.
واژگان شبکه مترجمین ایران