Logistic Regression Model Optimization and Case Analysis

Abstract

Traditional logistic regression analysis is widely used in the binary classification problem, but it has many iterations and it takes a long time to train large amounts of data, which is not applicable. In this paper, we study the mathematical model of logistic, define the error function, find the regression coefficient by gradient descent method, and improve the Sigmiod function. Finally, the number of iterations is reduced, and the classification effect is better, and the accuracy is basically unchanged. Besides, this paper establishes a vehicle evaluation prediction model to predict whether consumers accept a certain car. It provides a certain reference for the binary classification problem.