Abstract
The purpose of this paper is to discuss the state of fluctuation for the transportation demand which varies with time and to develop a time series prediction system and model considering properties of the fluctuation. The traditional models in time series analysis of data have problems such as insufficient consideration on the trend variation, complexity with including the seasonal variation and ambiguous process of model fitting.
A general prediction system and a new stochastic prediction model in time series analysis, AROP model, is proposed. Prediction system consists of subsystems to separate stochastic stationary or nonstationary process, to select a prediction model and to evaluate the precision of prediction. AROP model is divided into two models, AROP 1 and AROP 2, corresponding to trend variation.
