Abstract:
|
Users of official statistics are often interested in the effect of extreme or unusual events when interpreting movements in time series. This interest extends to the effect of the weather, for example, how extreme snow affected the UK's Gross Domestic Product (GDP) or how a warm autumn affected retail sales. The Office for National Statistics have been working in collaboration with the UK's producer of weather and climate statistics, the Met Office, and other UK government departments to analyse the effects of weather on a range of official statistics time series. Typically the highest frequency official statistics time series are monthly, so initial research used traditional time series methods on monthly retail sales, road accidents and ambulance response time statistics. One limitation of this approach is that weather effects could be obscured over a month and analysing higher frequency data could provide a better insight. Weather data are available at a daily and even hourly basis, as are road accident statistics. This paper presents results of analysing daily road accidents for weather effects and discusses the challenges of working with such high frequency time series.
|