The Business and Industrial Section of the RSS organised a session at the 2021 Conference in Manchester entitled 'Knowledge Transfer – Collaborations between academia and business through knowledge transfer partnerships (KTPs)'.
A KTP is a partnership between a business and a university which involves a novel business project being tackled with input from academics, to transfer knowledge to the business world. Each KTP is co-funded by the business involved and Innovate UK, a non-departmental body associated with the UK government. Projects typically last between 12 and 36 months and involve recruiting an 'associate' who is formally employed by the university but often works embedded in the business. At the conference session, we were fortunate to hear from three KTP associates whose projects included statistical aspects.
Speaking first was Ahmed Khafaga from Coventry University and Sencon UK Ltd. His talk was on 'Measurement uncertainty in the food and beverage metal packaging industry' and concerned the manufacture of aluminium ring-pull beverage cans. To assess quality during the manufacturing process, measuring instruments are used: using contact techniques during the manufacture and using optical techniques on the finished product, each with different calibration systems. The project on which Ahmed has been working involves understanding the uncertainties in each part of the process and thus producing a unified quantification of the measurement uncertainty.
Our second speaker was Ying Liu of the University of Leicester and Tangi0 Ltd, speaking about a 'Gesture recognition system using machine learning'. She gave some details of the etee device which is a button and wire-free handheld controller that can be used in virtual/artificial reality settings. It senses finger movements, touch and pressure to feedback on gestures being made by the user. Ying explained how machine learning techniques, including convolutional neural networks, are used to recognise data from the etee as representing different types of gesture. It can adapt to the diverse inputs created by the hands of different people and has applications in gaming and elsewhere.
Finally, Aedan Beatty of Queen’s University Belfast and CARD Group spoke about 'The application of choice modelling techniques to quantify subconscious priorities in the decision-making process'. In this project, discrete choice experiments (DCEs) were carried out where consumers are given different sets of options concerning a product or service and asked to choose which they prefer, with the data being analysed using discrete choice modelling. Using data from CARD, Aedan described the use of a multinomial logit model and how this could be improved by using a mixed logit model (to account for individual opinions and also include individual-specific coefficients) and spoke about the use of latent class models.
Author
Neil Spencer is vice-chair of the RSS Business and Industrial Section, professor of applied statistics at the University of Hertfordshire and director of its Statistical Services and Consultancy Unit