本文基于MCMC方法的核心思想，提出了一种通过Metropolis-Hastings 算法构造马氏链来估计变点序列后验分布的方法，从而检测变点出现的位置。 更进一步，本文在运行Metropolis-Hastings 算法时尝试将“温度”参数控 制了于一个较低的水平，从而得到了更好的估计结果。最后，我们将该算法 应用于一组真实气候数据，成功的找出了该序列的变点进而对该序列进行了 分段。

Based on the core idea of MCMC method, we propose a method, which constructs Markov chain using Metropolis-Hastings algorithm, to estimate the posterior distribution of change-point sequence, so that we are able to detect the position of change-point. In addition, we try to control the “temperature” parameter in a low level and our estimation result becomes better. At last, we apply the method in a data set of real-world climate. We successfully find out the position of change-point and divide the sequence into three parts.