Improving Predicting Accuracy by the Method of Least Squares

Abstract

Several predicting methods are analyzed in this paper, including: least squares method, extrapolation method, exponential smoothing method, adaptive smoothing method, mathematical modeling method, band method, matrix method, simulation method. One of these methods is aimed at solving the given problem using the methods of least squares linear regression and second order parabola method (quadratic regression). Currently, issues related to the approximation of points in the plane for calculating the coefficients of the function are relevant. Technologists, designers, and mathematicians use various approximation methods to average any number of measurements. Also in this paper, the analysis and programming of the above methods are aimed at finding the optimal function coefficients and helping the future application of these methods in the analysis of experimental data. In addition, a statistical approach aimed at modeling and predicting the level of fire risk based on the data coming from devices and environmental factors using the method of least squares has been developed.