关键词: 众数回归模型, 遗传算法, 模拟退火算法, 收入

Abstract: The classical mode regression model is mainly used to describe the relationship between the conditional mode of dependent variable and the independent variable, which is an important supplement to the mean regression and the quantile regression model. In this paper, we propose the optimization algorithms based on the genetic algorithm and simulated annealing algorithm to solve the parameter estimation vector of the classical mode regression model. The corresponding statistical test method is given, which makes up the lack of significant test method due to the absence of asymptotic theory for the classical mode regression model. In the empirical analysis, based on the Chinese General Social Survey data in 2010, this paper studies the impact factor of urban residents' income using the mode regression. The classical mode regression results indicate that the rate of return to education based on mode regression is 3.3%, far lower than the rate of return to education based on the mean regression and median regression, which is about 10%; the extreme value of experience is about 11 years based on the mode regression, far below the extreme value of experience based on the mean regression and median regression, which is 22 years more; for the largest proportion of urban residents, the income of labor contractors is about 15% higher than that of those who have not signed a labor contract. Based on these results, policy recommendations on narrowing the gap between the rich and the poor are given from education, technical training, and implementation of the law.

Key words: Mode Regression model, Genetic Algorithm, Simulated Annealing Algorithm, Income