# Introduction
Road safety is a complex phenomenon and needs several fields of study (urban planning, mathematics, road traffic management, smart cities). Road safety is part of the urban development agenda, also addressed in the UN Sustainable Development Goals [[arianeRoadSafetyEU]]. Comparing countries by following the trends of road fatalities plays an essential role in setting up road safety goals. Furthermore, the general UN goals have also been used as a basis for national strategic goals (Commission, 2018; Golinska & Hajdul, 2012).  Road traffic crashes cost most countries 3% of their gross domestic product (WHO, 2022).
Road safety has become a field of scientific study (Jamroz, 2008), as several theories and models have studied road safety. The article aims to explain why accidents happen from the random events theory perspective. Therefore, the research question was posed: how to measure road safety across countries? The authors provide a model for estimating road safety based on the number of fatal road accidents. This work aims to simplify the reality of road transportation, showing that the best predictor is the number of fatal road accidents, not the number of fatalities that have been often used in previous models. The reason is that the number of accidents follows a Poisson distribution to predict how many fatal accidents will occur per year (Spiegelhalter & Barnett, 2009).
Several definitions of road traffic accidents have been used in the literature (OECD, 2014). In this article,  two conceptual definitions have been used: (1) fatality rate as road deaths per  million of the population; (2) fatal accident as an accident in which a person involved in the accident has died within 30 days as a consequence of the accident (Finnish Data). This analysis emphasizes that the number of fatal accidents is a better predictor for road safety than the number of deaths on roads, taking into account that there are many accidents with multiple fatalities in one accident.  
Therefore, it is essential to consider multiple fatalities taken as one case; for example, there are fatal road accidents where two and more people are killed. The authors show that the daily fatal accidents follow the Poisson distribution (Spiegelhalter & Barnett, 2009), applying also the statistical significance test while comparing the countries (Benjamin et al., 2018; Shaver, 1993). The research identified that comparing road safety across big and small countries is challenging. In smaller countries the fatality rate is higher than in large countries.Therefore, the road safety measurement and prediction is especially important in smaller countries with higher number of fatalities. Thus, road traffic safety deserves special attention for sustainability purposes in modern urban ecosystems and science. This is because of high costs due to social and economic losses.  The model is provided to estimate the daily fatal accidents based on previous periods' data, depending on the country's size and the statistical significance of annual change. 
This paper focuses on explorative data analysis for fatal road accidents. The goal is to propose a prediction model for countries with different size. The authors´contribution to the road safety literature is to empirically predict the number of fatal accidents as this matters a lot for reducing deaths on the roads. In the previous literature, the models aim to predict the number of road accidents taking into account among other factors also the severity of accidents, casualty count and the number of vehicles involved (Gatarić et al., 2023; Pourroostaei Ardakani et al., 2023)
The key contributions of this research are: (1) to study the patterns of road fatal accidents; (2) to provide a best fit model for predicting road fatal accidents; (3) to contribute to the random events road safety theory.
The overall outline of the paper is following: section 2 reviews the theoretical background of the road safety research; section 3 focuses on materials and methodology and explains the data patterns. Secrtion 4 outlines the results and the key findings and section 5 concludes the empirical paper and discusses on the possibilities of the future research.