Open AccessModelling New Zealand Road Deaths
Author(s) -
Colin Morrison,
Ernest Albuquerque
Publication year - 2021
Publication title -
journal of road safety
Language(s) - English
Resource type - Journals
eISSN - 2652-4252
pISSN - 2652-4260
DOI - 10.33492/jrs-d-19-00246
Subject(s) - autoregressive integrated moving average , distributed lag , context (archaeology) , baseline (sea) , time series , autoregressive model , computer science , econometrics , macro , psychological intervention , operations research , geography , engineering , economics , machine learning , psychology , oceanography , archaeology , psychiatry , geology , programming language
New Zealand is developing an integrated road safety intervention logic model. This paper describes a core component of this wider strategic research carried out in 2018: a baseline model that extrapolates New Zealand road deaths to 2025.The baseline will provide context to what Waka Kotahi NZ Transport Agency is trying to achieve. It offers a way of understanding what impact interventions have in acting with and against external influences affecting road deaths and serious trauma.The baseline model considers autonomous change at a macro level given social and economic factors that influence road deaths. Identifying and testing relationships and modelling these explanatory variables clarifies the effect of interventions.Time-series forecasting begins by carefully collecting and rigorously analysing sequences of discrete-time data, then developing an appropriate model to describe the inherent structure of the series. Successful time-series forecasting depends on fitting an appropriate model to the underlying time-series.Several time-series models were investigated in understanding road deaths in the New Zealand context. In the final modelling an autoregressive integrated moving average (ARIMA) model and two differing autoregressive distributed lag (ARDL) models were developed. A preferred model was identified. This ARDL model was used to project road deaths to 2025.