George Botzoris

Econometric models for the forecast of passenger demand in Greece

Abstract The structure of models of forecast of passenger demand is first discussed. Parameters affecting modal split and mobility are analyzed. After many trial and errors procedures, three econometric models have been developed for the forecast of passenger demand in Greece: one for total demand, one for rail demand and one for private car demand. The validity of each model is tested by means of statistical and diagnostic tests, which are: Estimation of Coefficient of Determination (R 2), Collinearity test of independent variables, Statistical test of the F -statistics, Statistical test of the standard error, Model function form test, First degree self-correlation test to residuals, Residuals correlation, Heteroscedasticity and normality test, Model stability test and Forecasting ability of the models through U-Theil Statistics. The validity of the models has been also corroborated through the study of elasticities of independent variables. Once checked the forecasting ability, the models can be used for the forecast of future demand and modal split for passenger demand in Greece.

Application of Fuzzy Sets for the Improvement of Routing Optimization Heuristic Algorithms

Abstract The determination of the optimal circular path has become widely known for its difficulty in producing a solution and for the numerous applications in the scope of organization and management of passenger and freight transport. It is a mathematical combinatorial optimization problem for which several deterministic and heuristic models have been developed in recent years, applicable to route organization issues, passenger and freight transport, storage and distribution of goods, waste collection, supply and control of terminals, as well as human resource management. Scope of the present paper is the development, with the use of fuzzy sets, of a practical, comprehensible and speedy heuristic algorithm for the improvement of the ability of the classical deterministic algorithms to identify optimum, symmetrical or non-symmetrical, circular route. The proposed fuzzy heuristic algorithm is compared to the corresponding deterministic ones, with regard to the deviation of the proposed solution from the best known solution and the complexity of the calculations needed to obtain this solution. It is shown that the use of fuzzy sets reduced up to 35% the deviation of the solution identified by the classical deterministic algorithms from the best known solution.

Forecast of tourism demand with the use of fuzzy and cointegration econometric techniques

Econometric models for the forecast of tourism demand are developed in this paper. In order to assess the long-run trends concerning principal tourism generating countries, the Johansens maximum likelihood techniques are applied. For a better assessment of the short-run trends, the estimated error correction terms are introduced to the first difference models to estimate the short-run relationships (Error Correction Models, ECMs). Based on the results given by the ECMs, fuzzy regression models are suggested and then compared in order to provide the forecasting ability of both techniques (Fuzzy and ECMs). Finally, for the evaluation of forecasting performance, the Theils Inequality Coefficient is applied.

Clustering Variables Based on Fuzzy Equivalence Relations

We develop a method of grouping (clustering) variables based on fuzzy equivalence relations. We first compute the pairwise relationship (correlation) matrix between the variables and transform the matrix into a fuzzy compatibility relation. Then a fuzzy equivalence relation is constructed by computing the transitive closure of the compatibility relation. Finally, by taking all appropriate α-cuts, we obtain a hierarchical type of variable clustering. As examples, we use the proposed method first as a variable clustering tool in a regression model and secondly as a new way of performing factor analysis.

MODELING QUEUEING SYSTEMS USING FUZZY ESTIMATORS

Queuing modeling is the mathematical approach for the analysis of waiting lines. The central problem in every queueing model is a trade-off decision: The manager must weigh the added cost of providing more rapid service against the inherent cost of waiting. In this paper, a new fuzzy approach to queueing modeling is presented in order to eliminate the disadvantages of point estimation and the relevant paradoxes when calculating the efficiency of a queue, in the case of unreliable data available. Both Poisson arrival rate and exponential service time are considered by using fuzzy estimators. The introduction of fuzzy estimators in performance measures of M/M/S queueing systems is presented to address the central estimation issue under uncertainty.