Semi-Parametric Regression Models for Analysis and Forecasting of Freight and Passenger Transportation Volumes
2009
Diāna Santalova

Defending
07.12.2009. 14:30, 1, Lomonosova street, building V, room 218, Riga, Latvia

Supervisor
Aleksandrs Andronovs

Reviewers
Peteris Balckars, Viktors Ļumkis, Nikolai Ushakov

The promotional work «Semiparametric Regression Models for Analysis and Forecasting of Freight and Passenger Transportation Volumes» has been worked out by Diana Santalova to obtain the scientific degree of Doctor of Science in Engineering. Scientific supervisor of the work is Dr. hab. sc. ing., professor Alexander Andronov. The objectives of the present work are different kind of transport freight and passenger transportations forecasting applying modern statistical methods. Nowadays is observed not only industrial decentralization all over the world, but also increasing of population mobility. As a consequence is a steady growth of number of passenger and freight transportations by all kind of transport. This fact gives evidence that chosen direction of research is perspective, and presented work is actual. The first direction of the research is group models development and verification for transportations volumes forecasting for the EU Member States. In this connection four tasks are solved: • forecasting of total air freight and mail transportations on the basis of parametric multivariate model; • estimation of correspondence matrix of passenger rail departures applying modified gravity model; • evaluation and forecasting of rail freight turnover on the basis of semiparametric single-index model; • analysis and forecasting of total air passenger transportations in conditions of incomplete data using SURE-model. The second direction is elaboration and estimation of semiparametric single-index models for analysis and forecasting of passenger rail departures for regions of Latvia. The new methods of models estimation are suggested, criteria for the created models quality verification are developed. Advantages of application of the semiparametric models in comparison with the classical methods of the linear regression are shown. The promotion work is written in English. The given work consists of 7 chapters. The bibliography includes 128 sources. There are 30 figures, 147 formulas and 60 tables to illustrate the conception of the carried out research. The promotion work contains 164 pages.


Keywords
single index model, SURE-model, multivariate regression, transport, forecasting

Santalova, Diāna. Semi-Parametric Regression Models for Analysis and Forecasting of Freight and Passenger Transportation Volumes. PhD Thesis. Rīga: [RTU], 2009. 164 p.

Publication language
English (en)
The Scientific Library of the Riga Technical University.
E-mail: uzzinas@rtu.lv; Phone: +371 28399196