Statistics contributes to society by developing quantitative methodologies that respond to real-world problems in a variety of subject areas.
The main focus of our research is on modelling spatial and temporal variations of financial and natural processes. We aim at developing refined methods for detecting underlying patterns of, and interconnections between, such behaviour. Since the amount of data involved is often large, modelling requires new methods to fit models to the data. To develop these methods, we utilize efficient computational methods and rely on assumptions which go beyond the classic Gaussian paradigm.
From financial data to predicting the weather
To mention only a few projects at the Department of Statistics, we work with economists to model interactions of financial data within countries and markets, to develop dynamic stochastic models for efficient spatial analysis of linkages in financial markets and to understand impacts of multi-scale macroeconomic variables on market risks.
We work with ecologists to predict local moose populations, to predict the unknown location of oil tankers when oil spillage is observed on the sea and with meteorologists to model patterns of daily rainfall during the monsoon, and the evolution of such patterns within each season. We model compositional time series with application to party preference polls and develop spatio-temporal models of hurricanes and weather extremes in the Caribbean, to support the work of maritime scientists.
We participate in an interdisciplinary project entitled Ageing Risks in Life Expectancy Studies. It aims at developing an individual health index and a new time scale, called 'health age'. Statistical analysis will be carried out together with experts from fields such as ageing, epidemiology, health economics and science, medicine, insurance and risk management. This broad collaboration allows us to study the profound impact of ageing on various aspects of our society.
Another project we are pursuing is a study of the impact of climate by modelling rainfall on the Indian peninsula, specifically with the aim of extracting rainfall patterns during the monsoon season. Finally, statistical methodology is applied to personalized medicine, where we attempt to predict siRNA (Small interfering RNA) in cells using fluorescein proteins. The end goal is to be able to turn on and off genes in the presence of severe diseases such as cancer.
In addition to all current projects, we have just started an interdisciplinary project, The Statistics of Entrepreneurship, which aims at modelling small business milieu. The goal is to use the underlying latent processes to further understand the dynamic of growth and survival of new firms.
The Department of Statistics also plans to add data science to the scope of its activities with topics such as Machine learning and Regularization methods.