Non‐parametric probabilistic forecasts of wind power: required properties and evaluation

• DOI: 10.1002/we.230
• Publication Date: 2007-05-27T04:12:16Z
• Authors: Pierre Pinson; Henrik Aa. Nielsen; Jan K. Møller; Henrik Madsen; George N. Karinio...
• Keywords: Probabilistic Forecasting; Quantile; Prediction interval; Point estimation; Wind Power Forecasting

Abstract Predictions of wind power production for horizons up to 48–72 h ahead comprise a highly valuable input the methods daily management or trading generation. Today, users predictions are not only provided with point predictions, which estimates conditional expectation generation each look‐ahead time, but also uncertainty given by probabilistic forecasts. In order avoid assumptions on shape predictive distributions, these produced from non‐parametric methods, and then take form single set quantile The required desirable properties such forecasts defined framework their evaluation is proposed. This applied evaluating quality two statistical producing full distributions power. These number nominal proportions spanning unit interval. relevance interest introduced discussed. Copyright © 2007 John Wiley & Sons, Ltd.