Hjem
Matematisk institutt

Ph.d-kandidat Håkon Otneim holder seminar i selvvalgt emne

Tema for seminaret er “Non-parametric conditional density estimation: A survey"

Hovedinnhold

Non-parametric conditional density estimation: A survey

The estimation of conditional quantities is ubiquitous in the practice of statistics. There is a rich literature on regression techniques, dependence modelling and causality, but at the heart of it all lies often a conditional density function. Indeed, in many cases it makes good sense to estimate the entire conditional density of a stochastic vector given the value of another stochastic vector, instead of, for example, only the conditional mean or the conditional variance. 

In this talk, some existing methods for the non-parametric estimation of conditional density functions will be presented. Instead of focusing on the details, we will rather look at the big picture, and identify a few basic classes of estimators, and possibly a "hole" in the fauna of methods, leaving the practitioner with few options in some easily imagined situations. Can we do better?

Examples with real and simulated data will be included.