Abstract

The methods of probability theory substantiate a brief procedure that allows us to express the parameters of the polynomial regression of conditional mathematical expectation through mixed statistical moments of a system of random variables. Examples of linear and quadratic regression are implemented. In the second case, consideration is limited to the situation when the probability density of a random argument is an even function. The result was obtained without cumbersome calculations, since it was obtained using not the initial statistical moments resulting from the least squares method, but mixed central moments reflecting the type of regression curve. It is shown that in the general case, taking into account the nonlinearity of the correlation dependence only strengthens the inequality, which is the criterion for the adequacy of the regression approximation. The convergence of such a procedure is confirmed if the conditional expectation is not essentially a polynomial.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call