Method for solving logistic regression analysis using trust region based on tangent single broken line method

Abstract

The invention provides a method for solving logistic regression analysis using a trust region based on a tangent single broken line method. The method comprises: step S1, using a linear regression fitting function and a Logistic function to establish a prediction fitting function; step S2, after a cost function is obtained by maximum likelihood estimation derivation, obtaining a solving target oflogistic regression analysis; step S3, establishing a trust region model subproblem; step S4, using a tangent single broken line method to solve the trust region model subproblem, to obtain an optimalnumerical solution. Compared with the prior art, the method for solving logistic regression analysis using a trust region based on a tangent single broken line method has more rapid local convergenceand more ideal global convergence relative to a common classical gradient descent method, and since a secondary model is used to solve a correction, decline of a target function is more effective than that using a linear search method. Secondly, an algorithm which is improved than a Dogleg method is combined, the algorithm is the tangent single broken line method, the Dogleg method being a conventional method solving a trust region subproblem, so that operation efficiency is further improved.

Description

technical field [0001] The invention relates to the field of regression analysis, in particular to a method for solving logistic regression analysis based on the trust region of the tangent single broken line method. Background technique [0002] Logistic regression analysis is a generalized linear regression analysis model, which is often used in data mining, automatic disease diagnosis, economic forecasting and other fields. Logistic regression is mainly used in epidemiology. The more common situation is to explore the risk factors of a certain disease and predict the probability of a certain disease according to the risk factors. [0003] The usual practice of logistic regression analysis is to first construct a cost function, and then solve it to find the optimal solution. If there is an analytical solution to this function, then the result can be obtained by derivation. However, in most cases, the function is difficult to be derived or the analytical solution cannot be...

AI Technical Summary

Problems solved by technology

If there is an analytical solution to this function, then the result can be obtained by derivation. However, in most cases, the function is difficult to derive or cannot obtain an analytical solution. In this case, an optimization method is needed.

In the past, people tended to choose the gradient descent method. This is an old and simple method, but its disadvantages are slow convergence, low efficiency, and sometimes not reaching the optimal solution.

Images