基于两次改进灰色加权马尔可夫模型的CPI预测

CPI Forecast Based on Two Improved Gray Weighted Markov Model

利用灰色加权马尔可夫预测模型对居民消费价格指数（CPI）进行研究，并在此基础上进行两次改进.首先利用无偏灰色加权马尔可夫模型进行改进，消除了原有模型固有的偏差；又利用递进转移概率矩阵进行二次改进，在预测时增加最新的预测数据，去掉距离最远的数据，最后结合2013.06~2014.03的 CPI数据对2014年4月及5月份的数据进行预测.结果表明：两次改进的灰色加权马尔可夫模型与一般灰色加权马尔可夫模型相比，平均相对误差约减少30%，效果较好.

Using gray weighted Markov prediction model,the consumer price index （CPI）was studied, and on this basis two improvements were carried on.First,unbiased gray weighted Markov model was used to improve,and the inherent deviation of the original model was eliminated.And secondary im-provements based on the progressive transition probability matrix increased the latest forecast data in forecasting,and removed the data in the furthest distance.Finally,the CPI data from June,2013 to March,2014 were used to forecast the data in April and May of 2014.The results show that the two im-proved gray weighted Markov model compared with the general weighted grey Markov model,the aver-age relative error is about 30% less,and the effect is better.