Linear Correlation Communication in Multidimensional Method of Point Distributions
Abstract
Introduction: Modern production tends to decrease the volume of samples during routine tests. The existing methods of finding the linear correlation coefficient do not provide the necessary precision of the results when the sample volume is small. Purpose: We analyze the possibilities to apply the Pearson correlation coefficient after the virtual growth of the initial data table using the multidimensional method of point distributions. Results: The study showed that the usage of virtual growth of the original data table based on the point distribution method allows you to avoid a wrong determination of the linear correlation communication as weak negative instead of weak positive when analyzing small-volume multidimensional samples. Besides, the proposed method allows you to narrow down the rating spread of Pearson correlation coefficient values which can be used to additionally assess the size of the linear correlation communication when creating mathematical models according to passive data at the early steps of the research. The novelty of this approach is using the information on each single small-sample implementation, relying on the knowledge about the types of one-dimensional random variable distribution laws. The formation of a virtually increased table (joining of values) is performed by the maximum level of the probability density. Practical relevance: The multidimensional method of pointed distributions can be applied to control technological processes in real time at the stage of accumulating the information about the studied object.Published
2016-12-19
How to Cite
Popukaylo, V. (2016). Linear Correlation Communication in Multidimensional Method of Point Distributions. Information and Control Systems, (6), 96-98. https://doi.org/10.15217/issn1684-8853.2016.6.96
Issue
Section
Control in social and economic systems