A model of data distribution based on texture analysis

Abstract
To eatlmate the number of tuplea aatlsfyrng a certain query, a data dlstrlbutlon model IS proposed The model 1s based on a discrete approxlmat?on of the data space and belongs to the class of nonparametrlc models Using texture analysis techmquea apphed to the multl dImensIonal data apace, It 1s proposed that a segmentation of thla apace be obtaIned as a means of obtalnlng a discrete approxlmatlon Thus the space IS divided Into a number of homogeneous regions which can be later queried to obtain good estlmatea of the size of the response set To obtain this segmentation, a new function to assess the homogeneity of a bit pattern IS proposed Test results performrd for thla function are presented to show the Inverse correlation between Its value and the resulting estimation errors

This publication has 7 references indexed in Scilit: