Jiří Procházka, Milan Bašta, Matej Čamaj, Samuel Flimmel, Milan Jantoš
Trend and Seasonality in Fatal Road Accidents in the U.S. States in 2006–2016
Číslo: 2/2018
Periodikum: Statistika
Klíčová slova: Road traffic accidents, yearly seasonality, long seasonal period, generalized linear model, model selection, cluster analysis
Pro získání musíte mít účet v Citace PRO.
Anotace:
Understanding the dynamics of the daily number of fatal road traffic accidents is important for local authorities,
police departments, healthcare facilities and insurance companies, enabling them to design preventive measures,
provide appropriate emergency service and care and reliably estimate traffic accident insurance costs. In the
present study, using the Fatality Analysis Reporting System provided by the U.S. National Highway Traffic
Safety Administration, we construct a daily time series of the number of accidents for each state of the United
States. We model the trend as well as yearly and weekly seasonality present in the time series and provide
respective trend and seasonality statistics. Differences in accident rates and yearly seasonality between states
were detected, clustering analysis being applied to identify clusters of states with similar yearly seasonality,
weekly seasonal patterns for different states proving to be about the same.
Zobrazit více »
police departments, healthcare facilities and insurance companies, enabling them to design preventive measures,
provide appropriate emergency service and care and reliably estimate traffic accident insurance costs. In the
present study, using the Fatality Analysis Reporting System provided by the U.S. National Highway Traffic
Safety Administration, we construct a daily time series of the number of accidents for each state of the United
States. We model the trend as well as yearly and weekly seasonality present in the time series and provide
respective trend and seasonality statistics. Differences in accident rates and yearly seasonality between states
were detected, clustering analysis being applied to identify clusters of states with similar yearly seasonality,
weekly seasonal patterns for different states proving to be about the same.