Car accidents: How much is due to external factors and conditions? A data science approach for the Portuguese road network

• Autor(a) principal: Lourenço, Ana Sofia Gamarro Figo
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10071/27095
• topic: Car accidents; Traffic; Victims; Spatial dependence; Moran's index; Acidentes; Tráfego; Vítima -- Victim; Dependência geo-espacial; Índice de Moran

The main objective of this research is to find out which and what weight external factors have in accidents and victims resulting from them. Within the variable accidents, all accidents that happened in Mainland Portugal in 2018 are counted, according to INE, as the victims include all victims, in Mainland Portugal in 2018 resulting from an accident. The data used was taken from several sources, namely, PORDATA, IPMA, INE, DGTerritório and Here. After collecting the data, the data was thoroughly analyzed and new variables were created with the help of QGIS and SPSS Statistics software, all of them organized by municipalities belonging to the country under study. After all the analysis and selection of variables with the Geoda software and the literature, different models were performed in order to draw conclusions about the selected variables. For this study, two different models were made, for accidents and victims (per 1000 meters and per 1000 inhabitants respectively), because these two variables (targets) didn’t have a strong linear correlation, presenting a value of 0.036 (Pearson correlation) since there was no relationship between the variables. In order to generalize to Portuguese road structures and to other countries with similar characteristics to Portugal, the bootstrap method was used as a simulation strategy, thus generating 300,000 new data. After evaluation the data, it was found that the external factors used in these models have an explanatory capacity of less than 50%, but spatial dependence is a key and very important factor in geospatial problems.