Tid: 28 oktoberber 2002 kl 1515-1600
Plats : Seminarierummet 3733, Institutionen för matematik, KTH, Lindstedts väg 25, plan 7. Karta!
Föredragshållare: Jonathan Wendin.
Titel: Estimation of the Spectral Measure and the Tail Dependence Coefficient for Regularly Varying Random Vectors. (Master thesis)
Sammanfattning: Multivariate regular variation is introduced as a tool for modelling heavy-tailed multivariate phenomena. Regularly varying random vectors are characterized in terms of a tail index and a spectral measure, where the latter can be interpreted as a measure of dependence for extreme events. Estimators of the spectral measure as well as of exceedance probabilities for improbable events are proposed. A final application concerns estimation of the tail dependence coefficient - a measure of asymptotical dependence that is of immediate interest to risk management.