Using the Characteristic of Sample Data Distribution Law in the Machine Learning Tasks

Abstract

The paper introduces a new characteristic of a random variable with a bounded support titled a distribution indicator, defined as the ratio of the squared support length to the random variable variance. It is shown that this characteristic of the whole family of random variables of the same distribution law type. Computer simulation of random time series signals and evaluation of the distribution indicator had demonstrated the possibility of detecting the distribution law of the scries observations by analyzing this characteristic only.