Abstract
A conventional grey Markov model will produce large errors when it's used for the prediction of a sequence with large volatility. These errors can lead to prediction failure. So it is necessary to improve the model. In order to reduce the volatility, the logarithm of the original sequence is firstly calculated, and then the minimum error is used to replace the relative error of the grey model. After that, a Markov chain is employed to revise the predicted value. Finally, the conventional and the improved models are respectively applied in the fitting and prediction of Jingdezhen ceramic industrial output from 2003 to 2011. The result of the experiments shows that the improved grey Markov model can more accurately reflect the change in the ceramic industrial output of Jingdezhen.