Analysis Of Financial Data Based On Neural Networks And Genetic Algorithms

The stock market is a complex system,the stock market research is a hot topic in the economic field.Neural network has a good nonlinear system fitting ability,but the use of neural network model analysis and forecasting stocks,it is difficult to give the appropriate variable selection criteria.Genetic algorithm based on Darwin's "survival of the fittest" theory,through a suitable fitness function of the "guidance",so that high-quality genes(quality individuals)to a larger probability of inheritance to the next generation to survive.Using this method for variable selection,we can optimize the variables that affect the stock price effectively,and effectively solve the problem of selecting the input layer variables of the neural network.The applicability function only considers the prediction error,the smaller the prediction error,the higher the fitness of the individual.However,when the prediction errors are the same or similar,we should prefer those individuals with few variables.Obviously,the applicability function does not solve this problem.Based on this idea,this paper presents a new applicability function.The new applicability function not only considers the prediction error,but also the number of variables.Based on the new fitness function of the "guidance" to get the quality of individuals,both to ensure good predictions,but also has fewer variables.In this paper,a single artificial neural network model(BPNN),principal component analysis and neural network ensemble model(PCA-BPNN),genetic algorithm and neural network ensemble model(GA-BPNN)are used in this paper.Improved genetic algorithm and neural network combination model(IGA-BPNN).The results show that the proposed method can effectively reduce the number of variables while guaranteeing the basic prediction accuracy.