A HMM-based Genetic Algorithm Research For Forecasting

With the development of artificial intelligence and electronic commerce, supply chain walks into people’s lives, and it attracts people’s attention constantly. Prediction is an important part in supply chain, especially predict the sale price. However, predicting sales price is also a very complex issue in supply chain; for traditional prediction algorithms can not meet their requirements. Currently, as the wider use of the hidden Markov model in predicting price and the evolutionary algorithm develops rapidly, the price-prediction in supply chain makes use of the hidden Markov model as a new research idea and based on the theory of evolution method.Firstly, this paper reports on the use of the Hidden Markov theory, then presents a forecasting model which combines the Hidden Markov model (HMM) with quantum genetic algorithm. Because the hidden Markov model depends on the initial parameters strongly when it used in forecasting,and if single used,the prediction accuracy is not high,so taking the prospect of the combined forecasting model into consideration, and introducing the think way of "dividing and dealing" which come from the co-evolution algorithms, the hidden Markov is combined with genetic algorithms, however, when taking the number of iterations into account, the genetic algorithm prone to converge slowly,precocious, and trapped into local optimum problem, whereas the quantum genetic algorithm could overcome this deficiencies, so this paper chooses it.Then, the Hidden Markov quantum genetic algorithm is applied to the stock market,the stock data is selected on the Yahoo! Finance website experimental verification,after analyzing the forecast results and compared with other forecasting methods, the combined predicting model derived in this paper is proved to be more accurate in prediction.Finally , in order to solve the prediction problems in supply chain, the Hidden Markov quantum genetic algorithm is applied to the TAC/SCM Supply game platform. By predicting sales price accurately in the race, the marketing strategy can be adjusted in time and, finally, get more demand orders then make great gains.