Research On Change Point Of Express Delivery Business By Bayesian Analysis Theory

With the development of economy, the mathematical statistical problem that gets more and more attention of researchers is change point, because it was not only widely used in industrial quality control in early years, but also it had a large number of applications in the financial, computer, medical, economic and other fields in recent times.Firstly, this paper introduces the research status of domestic and foreign scholars in the field of the change point and the significance of this article. And secondly, it introduces the single change point and multiple change point. Thirdly, this paper is based on the previous studies, uses the method of change point of uniform distribute and probability to analysis the quantity of expressage in 2014 in Nanning City, from TAN to NOV.When using the change point of uniform distribute to research the quantity of expressage, which is based on the Bayesian analysis theory, first we should derive the posterior distribution by using the prior distribution, then use the maximum likelihood method to derive estimate the position of change point, and finally use what we get for the case analysis. When we use the change point of probability to analyze the quantity of expressage, It is reasonable that assume the quantity of expressage in several months before a year is a sequence of independent random variables and subject to distribution, while the quantity of expressage in the latter several months is independent random variables and follows another distribution, at this time the change point of express quantity can be detection, at last using this method to analysis the quantity of expressage.Through the research of the quantity of expressage of Nanning by using the two kinds of the methods of change point,it shows that the change point is really existence and can detect the time when the change point appears. By analysing and comprising, combined with the case study, it is true that the change point of probability method in the case of small samples is more accurate and consistent than the change point of uniform distribute method.