Comparison Of Several Time Series Models In Passenger Flow Prediction

Nowadays, forecasting is getting more and more popular in academic field. Obtaining data is easy in contemporary society. How to use it correctly and effective becomes a problem. Time series is one of the most popular methods to forecast the data in future. It is widely used in many fields such as economic field, commercial field and so on. Because it requires little background knowledge of the field you predicted. That is the reason time series develops so rapidly in so many fields. In this paper, several kinds of time series models will be used to predict the passenger flow. And it can predict the data in the near future with doing research on time series only. In this paper, there uses two existed time series models-fuzzy time series model and seasonal level model, a hybrid forecast model which combines the fuzzy time series model and seasonal level model to predict the passenger flow with the data of shopping arcade in Nanjing.The paper gives a detailed introduction of predicted process of three time series models which are fuzzy time series model, seasonal level model and hybrid forecast model. It finds out that the prediction of hybrid forecast model is best, the fuzzy time series model’s is worst and the seasonal level model’s is in the middle of three by comparing the prediction of three time series model. Because hybrid forecast model and seasonal level both take into account of the periodicity of time series. The data provided by this paper is high cyclic. Considering the character of data before forecasting is helpful to choose a suitable model. According to the character of the data, there comes up with the hybrid forecast model to predict the passenger flow of shopping arcade. The experiment result indicates that the new time series have a better accuracy.