Talks and Poster Presentations (without Proceedings-Entry):
E. Bura:
"Sufficient dimension reduction in econometrics";
Talk: New Developments in Econometrics and Time Series,
Graz (invited);
2019-06-06
- 2019-06-07.
English abstract:
Sufficient Dimension Reduction (SDR) summarizes a vector of predictors x as it relates to a univariate or multivariate response y, so that all the information in the conditional distribution of y|x is preserved. SDR appeared in the early 90īs and its methodology has grown significantly since then. SDR primarily comprises of moment based and model (distribution) methods for the inverse predictors. It encompasses both linear and non-linear reductions of the predictors and it is exhaustive. Nevertheless, SDR was developed for cross-sectional data. A general SDR framework for macro-forecasting and a comparison with widely used methods for analyzing large panels of macro-variables will be presented.
Created from the Publication Database of the Vienna University of Technology.