Data analytics and modelling

Developing techniques to improve modelling and analytics, enabling predictions and decisions to be made with confidence

Measurement provides data that is used as the basis for drawing inferences, informing decisions and controlling systems. These activities depend on the reliability and traceability of data and on the inter-operability of measuring systems. These characteristics are established through theoretical analysis, modelling and data analytics.

In the Data Science group we are building the capability to ensure that information derived from data is as reliable as possible and the quality of that information is quantified. We are developing techniques for optimisation and sampling to enable reliable inference under uncertainty, particularly in the context of large-scale problems. We are extending our work in measurement uncertainty evaluation to cover new methods of data analytics, such as machine-learning and classification tasks. We are building on our work in software testing to ensure that the algorithms and software used to process data can be demonstrated as trustworthy. Our capability in these areas will be applied across various sectors to provide case studies, demonstrators and facilitate external uptake.

Our work is focused on the following themes:

• Inference covering techniques for optimisation and sampling from probability distributions to support reliable inference under uncertainty
• New data analytics to extend our capabilities in experimental data analysis to cover statistical learning, machine learning and clustering for feature identification and classification
• Uncertainty quantification for data analytics, such as for machine learning and classification
• Software verification and validation, covering the formal definition of computational aims, generation of reference data, the design of performance metrics and the self-verification of software

Contact us