Maximizing the number of allowable approximation errors when building a linear regression model

Subject of research: errors in approximation of a linear regression model within the framework of a logical-algebraic approach to data analysis. Purpose of the study: to develop an algorithmic method for solving the problem of maximizing the number of permissible approximation errors using a linear-Boolean programming computer. Methods and objects of research: the object of research is a linear regression model, the methods are linear regression analysis and mathematical programming apparatus. Main results of the study: an algorithmic method is proposed for maximizing the number of permissible absolute and relative errors in approximation of a linear regression equation, which reduces to solving linear-Boolean programming problems of a dimension acceptable for practical situations. Solving generated problems of this type should not cause computational problems due to a significant number of effective software tools, for example, the LPsolve program, which is freely available on the Internet.