Advanced

Kör försiktigt - det är högkonjunktur

Kleverman, Marcus LU and Edin, Daniel LU (2017) STAH11 20162
Department of Statistics
Abstract
The purpose of this paper is to investigate whether a change of trend for traffic fatalities has occurred in the USA. Specific variables that are suspected to have an impact on fatalities are also analyzed. The idea for this paper came from a general ”call to action” from the White House to analyze the recent ”change of trends” for traffic fatalities in 2015. The data for which the analysis is based on comes from The National Highway Traffic Safety Administration. After a basic visual analysis, a possible correlation is noted between the level of unemployment and the level of traffic fatalities. Two types of analyses are performed with time series analysis and generalized linear models with the assumption of poisson distribution. Cross... (More)
The purpose of this paper is to investigate whether a change of trend for traffic fatalities has occurred in the USA. Specific variables that are suspected to have an impact on fatalities are also analyzed. The idea for this paper came from a general ”call to action” from the White House to analyze the recent ”change of trends” for traffic fatalities in 2015. The data for which the analysis is based on comes from The National Highway Traffic Safety Administration. After a basic visual analysis, a possible correlation is noted between the level of unemployment and the level of traffic fatalities. Two types of analyses are performed with time series analysis and generalized linear models with the assumption of poisson distribution. Cross correlation analysis of both time series is performed to assert the correlation. Poisson regression is used to detect which variables that have a significant impact on the number of fatalities. Nine states, divided into three stratums, are used in the poisson regression and these are chosen on the basis of their respective position in regard to the expected number of fatalities in the USA. An auto-regressive model of the second order describes the time series well. A significant negative correlation between unemployment and traffic fatalities is detected and the traditional variables within traffic safety such as alcohol consumption and speeding are significant. “The change of trend” is probably caused by a natural shift in the economic cycle that is mirrored by unemployment. By focusing more economic and political willpower on the identified variables, the number of traffic fatalities may be reduced. (Less)
Abstract (Swedish)
Syftet med denna uppsats är att undersöka om ett ”trendbrott” för antalet omkomna i trafiken i USA har skett. Utöver detta undersöks vilka specifika variabler som påverkar antalet omkomna signifikant. Uppslaget till detta arbete kom via en allmän uppmaning från vita huset i USA om att granska och analysera anledning till 2015 års ökning av omkomna i trafiken. De data som analyserna bygger på är framförallt hämtad från National Highway Traffic Safety Administration. Efter övergripande visuell analys utkristalliseras ett samband mellan arbetslöshet och omkomna i trafiken. Två typer av analyser genomförs med tidsserieanalys och generella linjära modeller med poisson ansats. Vidare bedrivs korskorrelationsanalys av tidsserier för att... (More)
Syftet med denna uppsats är att undersöka om ett ”trendbrott” för antalet omkomna i trafiken i USA har skett. Utöver detta undersöks vilka specifika variabler som påverkar antalet omkomna signifikant. Uppslaget till detta arbete kom via en allmän uppmaning från vita huset i USA om att granska och analysera anledning till 2015 års ökning av omkomna i trafiken. De data som analyserna bygger på är framförallt hämtad från National Highway Traffic Safety Administration. Efter övergripande visuell analys utkristalliseras ett samband mellan arbetslöshet och omkomna i trafiken. Två typer av analyser genomförs med tidsserieanalys och generella linjära modeller med poisson ansats. Vidare bedrivs korskorrelationsanalys av tidsserier för att kontrollera sambandet. Poissonregression används för att detektera vilka ingående variabler som har en signifikant påverkan på antalet omkomna. Nio stater, uppdelade i tre stratum, används i poissonregressionen och dessa väljs för sina respektive positioner i olycksstatistiken. Modellspecificeringen pekar på att autoregressiva modeller av andra ordningen beskriver tidsserierna väl samt vilka variabler som påverkar i de olika staterna. Ett signifikant negativt samband mellan arbetslöshet och omkomna i trafiken existerar och de variabler som signifikant påverkar omkomna i trafiken är de traditionella variablerna såsom alkoholkonsumtion och fortkörningar. ”Trendbrottet” beror med stor sannolikhet på en naturlig konjunkturell svängning som återspeglas i arbetslöshetsnivåerna. Genom att fokusera mer ekonomisk och politisk kraft på identifierade variabler så kan antalet omkomna i trafiken minimeras. (Less)
Please use this url to cite or link to this publication:
author
Kleverman, Marcus LU and Edin, Daniel LU
supervisor
organization
course
STAH11 20162
year
type
M2 - Bachelor Degree
subject
language
Swedish
id
8902264
date added to LUP
2017-02-28 12:19:58
date last changed
2017-02-28 12:19:58
@misc{8902264,
  abstract     = {The purpose of this paper is to investigate whether a change of trend for traffic fatalities has occurred in the USA. Specific variables that are suspected to have an impact on fatalities are also analyzed. The idea for this paper came from a general ”call to action” from the White House to analyze the recent ”change of trends” for traffic fatalities in 2015. The data for which the analysis is based on comes from The National Highway Traffic Safety Administration. After a basic visual analysis, a possible correlation is noted between the level of unemployment and the level of traffic fatalities. Two types of analyses are performed with time series analysis and generalized linear models with the assumption of poisson distribution. Cross correlation analysis of both time series is performed to assert the correlation. Poisson regression is used to detect which variables that have a significant impact on the number of fatalities. Nine states, divided into three stratums, are used in the poisson regression and these are chosen on the basis of their respective position in regard to the expected number of fatalities in the USA. An auto-regressive model of the second order describes the time series well. A significant negative correlation between unemployment and traffic fatalities is detected and the traditional variables within traffic safety such as alcohol consumption and speeding are significant. “The change of trend” is probably caused by a natural shift in the economic cycle that is mirrored by unemployment. By focusing more economic and political willpower on the identified variables, the number of traffic fatalities may be reduced.},
  author       = {Kleverman, Marcus and Edin, Daniel},
  language     = {swe},
  note         = {Student Paper},
  title        = {Kör försiktigt - det är högkonjunktur},
  year         = {2017},
}