To the question about removing errors in the process of mathematical modeling of metallurgical processes within the regression data analysis using the MathCAD package

This article presents frequently used methods for assessing the quality of regression models and their application in the study of certain metallurgical processes. These models are: R2 coefficient, correlation coefficient, multicollinearity, extrapolation, time systems, engineering prediction, full factorial experiment.Until now, a significant part of researchers have indicated that coefficient is used not only by economists, but also by graduate students of technical specialties as a diagnostic value, while the correction of this coefficient does little. The correlation coefficient is a numerical characteristic of the joint distribution of two random variables, independent of the dimension of the units. This rule applies equally to the correlation coefficient obtained by multiplying two matrix-columns. The method of data processing in the presence of collinearity (intercorrelation) in them and its elimination is shown. The drawbacks of using the extrapolation method in the process of mathematical modeling are shown. A new forecasting method has been developed, based on finite time series, called the sliding matrix method and consisting in the continuous updating of the coefficients of the regression equation by removing the line with obsolete data from the matrix and entering new lines with data at the predicted point. The method allows you to continuously get rid of the old information “burden”, which allows you to make the forecast more correct. All calculations of the mathematical model were made using the Mathcad software product.