The Research Of Products Recovery Prediction Based On The Least Squares-Markov Chain Model

With the progress of society and development of science and technology, ecologicalbalance is destroyed, energy is facing exhaustion and environmental problem is paid moreand more attention. Facing these problems, many solutions are proposed. China also putforward the concept of scientific development from national strategy level, emphasizingharmonious development of people and society, people and nature.The uncertainty of reverse logistics is a problem in practice which starts from thereverse logistics appearing. Because of the uncertainty, it’s difficult to do forecast work ofreverse logistics which is the basis for reverse logistics work carrying out in practice.Because of the reason, the paper on the basis of previous studies analyzed thecharacteristics of reverse logistics, found uncertain factors influencing reverse logistics,concerned the key factor and gave improved method, in order to make reverse logisticssteady as far as possible and improve the accuracy of prediction. On the basis, combinewith the advantages of Least Square Forecast and Markov Forecasting and use LeastSquare-Markov Chain model to forecast the reverse logistics products recovery.The paper has got the progress above:1) Reverse logistics influence factors analyses: The paper is based on AnalyticHierarchy Progress (AHP), giving out influence factors and sorting with weight. Inaddition, the paper gives corresponding measures for two of the biggest factors.2) This paper solved the problem of reverse logistics uncertainty and made it steady.Least Square Forecast model has the features of little data, small fluctuation, and suitablefor short-term trend prediction. Markov model in prediction can better solve the problemof random fluctuations. Taking advantage of the two models, we got Least Square-MarkovChain model which makes reverse logistics products recovery prediction relatively moreaccurate.3) Taking waste automobiles recovery of a company in Hebei as an example, LeastSquare Forecast model, Markov model and Least Square-Markov Chain model are used todo forecast work. And the paper gets the conclusion that Least Square-Markov ChainForecast model having great superiority through comparison.