This workshop focuses on the statistical analysis of biological and environmental data from a multivariate point of view. This means that we are not satisfied simply to summarize single variables in isolation - we wish rather to explore relationships between and within groups of biotic and abiotic variables in order to characterize the ecological system that is operating. The first part of the workshop introduces statistical techniques that describe and model biological variation and environmental variation separately. The second part is devoted to relationships between biological and environmental variables, using distance-based graphical methods such as canonical correspondence analysis. By the end of the workshop participants should be able to tackle advanced statistical analyses of their own data.

This workshop is aimed at post-graduate students and researchers in all branches of environmental science as well as statisticians who are interested to enter this area of application.

In the real world, when information only comes in through data and 'true laws' are not available, mathematical modeling may be an alternative to express the analyst's view. Naturally, such models need to be fitted to the data. In doing so, there may be a considerable risk that the proposed model is inadequate. Hence fitting a model should always be accompanied by a model check. For the simple multinomial model, with the chi-squares test, things started more than one hundred years ago. Decades later, Kolmogorov and Smirnov, Cramér and von Mises and others studied goodness-of-fit procedures for continuous data. These methods incorporated, for the first time, the concept of stochastic processes. With a deeper knowledge and understanding of stochastic processes, in the last 30 years, there has been a considerable progress in model checking techniques.
It is the goal of the workshop to present relevant ideas in model diagnosis in, e.g., the i.i.d. case, regression, time series, survival analysis and some selected semiparametric models.

The presentation will be such that both researchers from academia and applicants of statistical methodology will (hopefully) benefit from the lectures.

Professor Robert Engle from the Stern School of Business.

This talk discusses the concept of risk in financial markets, how it is measured with volatility models and how it looks today. It then focuses on the causes of the recent financial crisis with a brief mention of the regulatory reforms that are being proposed for its solution. The long run risks facing our society, including global overheating are addressed. It then goes on to address the measurement of risk in financial markets and how the use of standard Value at Risk measures ignore the risk that the risk will change. A proposal of additional risk measures and a discussion of portfolio strategies follow as well as a demonstration of how the use of short term risk measures contributed to the financial crisis