The Department of Statistics conducts research in several areas of statistics. These areas are presented below. Please visit the researchers´ personal presentations for more detailed information.
Statistics is an integral part of modern data science that provides quantitative methodologies for inference from data subject to uncertainty.
The strength of our research is both in methodological development and in applications. On the methodological side we have active research in spatial and temporal stochastic modelling, high dimensional analysis, regularisation methods, Machine Learning, extreme value theory, and compositional data analysis. Those methods are applied in a variety of disciplines including environmental sciences, finance and economics, mechanical engineering, public health, among others.
From infectious diseases to political opinion polls
The geographical distribution and emergence of infectious diseases are highly responsive to climate change. In one of our current research projects, we aim at developing robust climate-based algorithms and decision-making frameworks to support public health adaptation to infectious disease risks following from climate change. For this purpose, AI technologies, such as data mining and machine learning, are utilized to provide new algorithms and prototype decision-making dashboards of value for public health protection, specifically tailored to the needs of the European Centre of Disease Prevention and Control.
The Department has a broad expertise in spatial and temporal modelling. For example, we develop statistical inference for spatial(-temporal) point processes, both from a theoretical and an applied point of view.
We contribute to entrepreneurship research by addressing the following questions:
1.How can we better model entrepreneurship when the business opportunity remains unobserved?
2.How can we better develop theories of entrepreneurship using Structural Equation Modeling?
3.How can we better model and understand new firm growth and survival?
Answering these questions allow us to provide better advice to policy makers about what they should focus their attention on: empowering people to become entrepreneurs, focus on subsets of high-quality entrepreneurs (high in social, financial, and human capital), support specific sectors or, support a vibrant small business sector.
In environment sciences, we developed statistical methods to capture spatial structures in compositional data using Bayesian hierarchical models, Gaussian Markov Random Fields method and Stochastic partial differential equations.
In modern genetic studies, we developed efficient methods for identifying predictors in large data bases, with the specific emphasis on identifying causal mutations and building predictive models based on Genome Wide Association Studies. Our results will expand the mathematical understanding of high dimensional statistics, and provide a bridge, through open-source software, between the theoretical results and the practitioners. The application has a potential to have a huge impact on many different fields like biology, psychology, social science, and many more.
The theory of random graphs is also represented within our research fields. By using methods of the probability theory, we study random graph geometry which is applicable to random geometry in higher dimensions.
Political opinion polls play an important role in Swedish political life, as well as in many other countries. Polls estimate vote shares, i.e., proportions. This is known as compositional data. By combining several polls, we can create accurate error-controlled predictions of election outcomes. In our research, we combine diverse methodologies and techniques to advance our understanding of compositional time series, and enable better prediction of election outcomes.
List of our research in statistics in the Lund University Research Portal
Read about our seminars