Implementations generally relate to discretization for big data analytics. In some implementations, a method includes receiving a plurality of data points. The method further includes determining a plurality of data centroids, where each data centroid of the plurality of data centroids is predetermined. The method further includes determining a plurality of data regions for the plurality of data centroids, where each data region of the plurality of data regions is associated with a respective data centroid of the plurality of data centroids. The method further includes generating a plurality of data region boundaries for the plurality of data regions based at least in part on one or more data points of the plurality of data points, where each data region boundary of the plurality data region boundaries is associated with a respective data region of the plurality of data regions.