Lower and Upper Covariance
We give a definition for lower and upper covariance in Walley’s theory of imprecise probabilities (or coherent lower previsions) that is direct, i.e., does not refer to credal sets. It generalizes Walley’s definition for lower and upper variance. Just like Walley’s definition of lower and upper variance, our definition for lower and upper covariance is compatible with the credal set approach; i.e., we also provide a covariance envelope theorem. Our approach mirrors the one taken by Walley: we first reformulate the calculation of a covariance as an optimization problem and then generalize this optimization problem to lower and upper previsions. We also briefly discuss the still unclear meaning of lower and upper (co)variances and mention some ideas about generalizations to other central moments.
KeywordsVariance Covariance Central moment Theory of imprecise probabilities Envelope theorem
Unable to display preview. Download preview PDF.
- 1.Couso, I., Dubois, D., Montes, S., Sánchez, L.: On various definitions of the variance of a fuzzy random variable. In: De Cooman, G., Vejnarová, J., Zaffalon, M. (eds.) Proceedings of the Fifth International Symposium on Imprecise Probability: Theories and Applications (ISIPTA 2007, Prague, Czech Republic), pp. 135–144. Action M Agency, SIPTA, Prague (2007)Google Scholar