Statistical learning and data science
Statistics is the science of learning from data. Such learning processes aim typically to either extract information about the data generating mechanism or to predict future data. In the former case, in which the learning process is generally referred to as statistical inference, knowledge is represented by the underlying probability distribution of the data; in the latter case, referred to as statistical learning, knowledge is represented by a predictive function generating the data from given input variables. In any case, probability theory provides the mathematical tools for modeling the noisy nature of the data as well as understanding any aspects related to the accuracy or evidence of the statistical analysis.
Statistics is playing a fundamental role in science and industry. Today computer-based technology allows scientists and actors in industry and society to collect enormous amounts of data, which has given rise to the field of data science where computer science and statistics meet. The research activities and course syllabus of our department reflect this development. More specifically, the division of Mathematical statistics within KTH Mathematics conducts research in a variety of data science directions such as
- Monte Carlo simulation and computational statistics
- multivariate statistics and complex dependencies
- statistical inference in dynamical systems
- Bayesian networks
- generative models