Robust methods for heteroscedastic regression models for time series

What is the variability of the exchange rate of Euro to Australian dollar? Can we improve the use of the
electrocardiogram of a patient as a diagnostic tool for heart disease? A well-known limitation of the existing
statistical methods for answering these types of questions is that a small proportion of extreme observations
have the potential to lead to results that are more in agreement with the outliers than with bulk of the data. As
a consequence, the statistical analyses may lead to wrong conclusions. This project aims to develop new
methodologies to solve this problem for a large class of studies. Applications to stock market risk, exchange
rate, and diagnosis of heart diseases will illustrate the new methods.