Dimitris Andreikos, “Spatial analysis of road accidents in the municipalities of Greece”, Diploma Thesis, NTUA, School of Civil Engineering, Athens, July 2024.

The objective of the present Diploma Thesis is the examination of the influence of the spatial factor in the analysis of road accidents, as well as the degree of improvement of the analysis methods with the use of a geographically weighted regression. To that end, population, social, economic and transportation data was collected, and its influence on the number of road fatalities was examined. Following an initial processing of the data, their clustering procedure produced two clusters: one with the urban, densely populated Municipalities, and one with the regional and rural Municipalities. Subsequently, and following many tests, six final mathematical models were developed: three linear regression models for all the Municipalities and for each cluster separately, as well as three geographically weighted regression models, respectively. The application and evaluation of the models showed that the most important factors that influence the number of road fatalities are the possession of private vehicles and the Gross Domestic Product. Private vehicle possession correlates positively with the number of road fatalities, as the increase in vehicles leads to an increased exposure to danger, while GDP correlates negatively, as its increase is linked to better life quality and improved road safety measures.
In conclusion, it appears that the spatial models produce improved results when compared to the classic linear regression, as demonstrated by the R2 and AIC indexes. Therefore, the application of a spatial method of analysis improves the accuracy of the mathematical model estimation.