M-estimators of location for interval-valued random elements

Among the new types of data emerging from real-life experiments, interval-valued ones are becoming very prevalent nowadays. In summarizing the location of interval-valued datasets, the Aumann mean is the most usual measure. This measure inherits almost all the nice properties of the mean value for real-valued datasets. Nevertheless, it also inherits a critical property, which is the one related to its high sensitivity to data changes or to the presence of outliers.As an approach to measure the location of interval-valued datasets in a more robust way, the notion of M-estimators will be considered. Two applications on chemical data will be included to motivate and illustrate the problem. Finally, an empirical comparative study will be conducted to show the performance of the different types of M-estimators proposed in this work.