Multinomial Logit Model And Its Application In Postgraduate Professional Selection Behavior

Logit regression model can be used to analyze multivariate data and predict discrete dependent variables.Its independent variables can be continuous or discrete,or both.For the observed data,if the dependent variable is discrete,its categories are usually three or more than three.At this time,the Logit regression model we used is turned into Multinomial Logit regression model.Since the Multinomial Logit model has been proposed,it provides a good method for analyzing the discrete data.This paper introduces the establishment method and parameter estimation method of Multinomial Logit and Mixed Multinomial Logit in discrete choice model,which use the method of maximum likelihood estimation in Multinomial Logit model,and introduced the Guass Quadrature in parameter estimation and Monte Carlo model estimation method of Mixed Multinomial Logit.The model using Wald test and likelihood ratio test to evaluate the significance of the model.At the same time,Pearson 2?,deviance,the information measures and the goodness of fit were introduced,and the fitting degree of the model was evaluated by them.Finally,the application of the theory to a university graduate selection behavior and its preliminary test result is analyzed and forecast.Combined with various evaluation methods and prediction of the correct rate,we can choose the best model for this example.