Research On Forecast Model And Application Of An Improved Fuzzy Markov Chain

This paper presents a Markov chain forecast model based on different historical data weighted fuzzy state transition probability matrix, whose weight optimization search method was presented applying genetic algorithm. The purpose of optimization is to minimize historical data error, therefore, the new model is better than the old one both on the accuracy of siding-to-siding block and on the diminutive of the forecasting error.In order to verify the validity of the model proposed, the paper applies it to the study of local revenue forecasting. Meanwhile, the model is compared with the traditional regression forecast model and unweighted fuzzy Markov chain forecast model. At last, it is concluded that the precision of weighted fuzzy Markov chain forecast model which the paper proposed is higher than other forecast models.At the same time, the paper elementarily establishes the monthly local revenue forecast model, so as to make the Markov chain forecast model applied in the larger-scale.