Development of the software package “interactive automated system for optimal regressions modeling”

Abstract

© SGEM2018. Since the global information networks are well developing and the amount of information is rapidly increasing, an automatization of the analytical data processing has become more urgent. In their combination a solid basis for the improvement of management and operation packages in any field of activity arises. It should be noted the initial assumptions of the regression analysis are always observed. However, discovering that the preconditions are violated is not sufficient. A specific software package containing particular measures that come into force under these conditions are required. Thus, for the effective use of adaptive regression modeling approach (ARM) one should the apply a particular software package to automate the process of taking observations, analyze the quality of the models produced and analyze the compliance with the assumptions of regression analysis using the ordinary least squares method (LSM), as well as implement the appropriate procedures to adapt. The purpose of this study is to improve the performance of the computational modeling process by automating the search for the optimal set of regressors, and analyze it. To achieve this goal, it is necessary to solve a number of problems: 1) Development of the software package "Interactive Automated System for Optimal Regressions Modeling" (IASORM) based on connecting library quality analysis model with the compliance status of assumptions; 2) Implementation of the algorithm scenario of automatic data processing with the functional connection of libraries. The software package IASORM is a specialized system that implements the strategy of regression modeling. Automated script processing can improve the effectiveness of the existing methods of the package. Embedded library of the quality analysis and of the compliance model assumptions extend functionality for the user and is aimed at identifying the adequacy of models and observations in order to detect violations of the basic assumptions of regression analysis. Proposed scenario increases the computational process speed compared to interactive computing. IASORS implements the strategy of statistical (regression) modeling. This software package can be used to create regression models and predict dynamic processes.