Normal Return Of Investment Prediction By Neural Network Based On Genetic Algorithm

With the rapid development of computer technology and its application in all sorts of fields, many new subjects were started. Typically, the overlapping and development between life science and engineering is a great example in modern science and technology. It is also a research hotspot of relative fields. There are still many issues in modern scientific theory and practice with regard to Combination & Optimization and data analysis etc. The routine methods are helpless for complicated large-scale systems. So the two new subjects were founded by simulating the genetic and evolution mechanism of biology and simulating the reactions of neural cell. They are Genetic Algorithm (G.A) and Artificial Neural Network (A.N.N).The applications of the two subjects are related to more and more fields. Further more, with the great searching ability, G.A can solve many problems that other algorithms can't do. G.A was combined with A.N.N, and improved the capability of the A.N.N greatly. The combination makes the Intelligent Computing develop rapidly. So, to study the two subjects is very significant.This paper starts with the basic theory of Genetic Algorithm. With the view that G.A is applied in A.N.N, a new improved algorithm that optimizes A.N.N was designed. With the improved algorithm, A.N.N was used in a new forecasting. The major tasks included in this paper:(1) To analyze the basic theory of G.A and B.P, and to research the combination of G.A and B.P.(2) To give out a new searching operator, with adopting crossover operator and mutation operator, it forms an improved algorithm, comparing with the traditional algorithm, it is more effective.(3) Using the A.N.N to forecast the normal return of investment. The forecasting model is set up based on the financial data. So this paper enlarges the applications of A.N.N. Meanwhile, with the pattern classification performance of A.N.N, the way to improve the model was presented.