Research and Application of Multiple Regression Analysis Based on Genetic Algorithm

This paper deals with parameters selection of multiple regression analysis by genetic algorithm. Though analyzing which parameters having relations with the output of aluminum in the aluminium electrolytic industry, we constructed multiple regression models. We used binary code for individual who is in the solution space, selected individual by fitness ratio, then reproduction, crossover and mutation for the population. With a large number of iterative, we got the approximate solution of equation's coefficients. The experimental results indicate that using genetic algorithms to solve the multiple regression equation coefficients with high precision and intelligent. This model is suitable for forecasting, and can be used to guide practice.