Data for: 'Missing data in wind farm time series: Properties and effect on forecasts'

  • Rosemary Tawn (Creator)
  • Jethro Browell (Supervisor)
  • Iain Dinwoodie (Supervisor)



Data comprises two sets of wind farms: Dataset A is 30 European wind farms, with missing data, and dataset B is 10 UK wind farms with no missing data which is then used to explore the effect of introducing missing points. The proportion and properties of missing data in dataset A are found, and 3 different instances of missing data are reproduced in dataset B to determine their detrimental effect on forecasts and test mitigation methods. The three instances of missing data tested are missing values in the historical training set, missing forecast inputs when generating a new 'live' forecasts, and short historical data at new sites. This data also relates to chapter 3 of thesis 'Predictive analytics for wind power forecasting'.

Access to data restricted due to commercial constraints.
Date made available16 May 2022
PublisherUniversity of Strathclyde
Temporal coverage1 Jan 2017 - 18 Mar 2019
Date of data productionJan 2019 - Mar 2020
Geographical coverageEurope

Cite this