Stochastic Analysis of free flow traffic with emphasis on long-term persistence

The scope of this thesis is the stochastic analysis of free flow traffic with emphasis on long-term persistence and the identification of the statistical characteristics of speed, volume and occupancy time series and finally the finding of a prediction model for the speed. The data were derived from the highway I-205 at Portland,U.S.A. and from the Tolls of Polymylos on Egnatia Motorway. By studying the results of distribution functions, climacograms and Q-Q Plot, the first-order autoregressivefirst- order moving average was chosen. In addition, the application of the Hurst- Kolmogorov model to the real world data showed that this type of model is not suitable. Moreover, a three-parameter transformation was used in order to approximate the distribution of the historical speed time series by the normal distribution. Finally, for the first-order autoregressive approach, synthetic speed time series were produced with Monte Carlo method and subsequently comparative charts were presented between synthetic and historical speed time series.