A0579
Title: Wasserstein distributional data analysis with application to wind forecasting
Authors: Matteo Pegoraro - Aalborg University (Denmark) [presenting]
Abstract: Many environmental data come in the form of probability distributions. Due to the uncertainties involved in environmental processes, data are often aggregated to compare different phenomena better. Also, predictions are better understood regarding probability distributions over the possible outcomes. It is thus very important to develop techniques which can be used to solve data analysis problems related to distributional data sets. A framework is presented for principal component analysis and regression when statistical units are probability measures considered with the Wasserstein metric.
