基于加权马尔可夫模型的个性化需求预测

Individual Demand Forecasting Based on Markov Chain Model with Weights

针对个性化需求随机性和自相关性的特点,论述了个性化需求预测中引入加权马尔可夫模型的必要性及可行性,阐明了预测的具体步骤:利用销售序列的均方差建立分级标准,以规范化的各阶自相关系数为权重,结合转移概率矩阵计算出预测时段的状态概率,并确定预测时段的销量区间,进一步分析加权马尔可夫模型的特征。个性化定制下的快速消费品销售预测的算例分析表明,加权马尔可夫模型适合个性化需求预测。

In view of the randomness and auto-correlation of individual demands, the paper discusses the nccessity and feasibility of intro- ducing the Markov chain model with weights into the forecasting of individual demands and elaborates on the specific procedures of the forecast- ing： first a classification standard is set up using the mean-square deviation of sales series, then the standardized self-coefficients are used as weights, and the state probability of the forecasted time period is obtained through the transition probability matrix, and then the sales interval of the forecasted time period is determined, and finally the characteristics of the Markov chain model with weights are analyzed. A case study with the fast moving consumer goods under instant customization verifies the suitability and effectiveness of the model in individual demand forecasting.