Cigarette Sales Prediction in China Based on Markov Model

Abstract

Based on the analysis of the cigarette sales in past years, this paper introduced an effective and relatively precise method to predict the cigarette sales in China. The latest cigarette data published by government was collected to apply Markov Model. A novel way of processing sales data by sales ratio was adopted in classifying the states. In the prediction table constructed by the transition probability matrix, weighted average method was utilized to quantitatively and accurately simulate the sales prediction. Finally, the relative error between predicted and real national cigarette sales for 2012 to 2015 are -0.56%, -0.84%, 3.11% and 4.39% respectively. The research results offer the reference directly to enterprise investment decision and the industry development strategy.