Community Research and Development Information Service - CORDIS

Methodologies for the statistical analysis of activity time series in order to extract the weather-related signal

The statistical analyses sought to model the impacts of weather events (the predictors) on activity time series (the predictands). The models used were regression models, in which the predictor weather variables were principally rainfall, temperature and sunshine and the predictands were activity measures of weather-sensitive sectors.
Time series of the activity measures were extracted and graphed, and any long-term trend was either removed or incorporated in the regression equations as a predictor variable. The assumption is that the residuals from the long-term trend include short-term fluctuations in activity due to anomalous weather conditions. Where the climate variables are significant predictors of the residuals, a measure of the impact on the sector can be extracted and assigned a value.

Reported by

University of East Anglia
University Plain
NR4 7TJ Norwich
United Kingdom
See on map
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top