This isn't exactly what you're talking about but I think that it might interest you: Adaptive Data Partitioning Using Probability Distribution by Xipeng Shen, Y

This isn't exactly what you're talking about but I think that it might interest you: Adaptive Data Partitioning Using Probability Distribution by Xipeng Shen, Yutao Zhong, and Chen Ding (http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.1.3117 http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.1.31...). They estimate the probability distribution using a subset of the data and then use this to determine the partitions.