Experience of logistic regression parameters calculation by Newton-Rafson method to estimation resistance to cold of plants
Vasiliev N.P., Egorov A.A.
Saint-Petersburg State Forest-Technical Academy, Saint-Petersburg, 194021, Russia
nikpv@mail.ru
Abstract. When calculating the parameters of the logistic regression, which was used as a statistical model to estimation resistance to cold of plants, the authors faced the problem of unstable calculations. As is known, the task of constructing a logistic regression is reduced to a nonlinear system of equations. Newton-Raphson method is used to solution of this system. For some versions of the original data the convergence of the approximations is not stable. The possible solutions to this problem are discussed. The final results are compared for a number of criteria with results obtained using the well-known statistical program SPSS.
Key words: statistical processing, logistic regression, nonlinear system of equations, method of Newton-Rafson, instability of calculations, resistance to cold of plants.