Closing the Energy Efficiency Gap: A study linking demographics with barriers to adopting energy efficiency measures in the home

This paper presents a study which linked demographic variables with barriers affecting the adoption of domestic energy efficiency measures in large UK cities. The aim was to better understand the 'Energy Efficiency Gap' and improve the effectiveness of future energy efficiency initiatives. The data for this study was collected from 198 general population interviews (1.5-10 min) carried out across multiple locations in Manchester and Cardiff. The demographic variables were statistically linked to the identified barriers using a modified chi-square test of association (first order Rao-Scott corrected to compensate for multiple response data), and the effect size was estimated with an odds-ratio test. The results revealed that strong associations exist between demographics and barriers, specifically for the following variables: sex; marital status; education level; type of dwelling; number of occupants in household; residence (rent/own); and location (Manchester/Cardiff). The results and recommendations were aimed at city policy makers, local councils, and members of the construction/retrofit industry who are all working to improve the energy efficiency of the domestic built environment. © 2012 Elsevier Ltd.

Evidence-informed development of engineering practical work: Effecting change through staff and student engagement

This paper discusses innovations in curriculum development in the Department of Engineering at the University of Cambridge as a participant in the Teaching for Learning Network (TFLN), a teaching and learning development initiative funded by the Cambridge-MIT Institute a pedagogic collaboration and brokerage network. A year-long research and development project investigated the practical experiences through which students traditionally explore engineering disciplines, apply and extend the knowledge gained in lectures and other settings, and begin to develop their professional expertise. The research project evaluated current practice in these sessions and developed an evidence-base to identify requirements for new activities, student support and staff development. The evidence collected included a novel student 'practice-value' survey highlighting effective practice and areas of concern, classroom observation of practicals, semi-structured interviews with staff, a student focus group and informal discussions with staff. Analysis of the data identified three potentially 'high-leverage' strategies for improvement: development of a more integrated teaching framework, within which practical work could be contextualised in relation to other learning; a more transparent and integrated conceptual framework where theory and practice were more closely linked; development of practical work more reflective of the complex problems facing professional engineers. This paper sets out key elements of the evidence collected and the changes that have been informed by this evidence and analysis, leading to the creation of a suite of integrated practical sessions carefully linked to other course elements and reinforcing central concepts in engineering, accompanied by a training and support programme for teaching staff.

Student-t processes as alternatives to Gaussian processes

We investigate the Student-t process as an alternative to the Gaussian process as a non-parametric prior over functions. We derive closed form expressions for the marginal likelihood and predictive distribution of a Student-t process, by integrating away an inverse Wishart process prior over the co-variance kernel of a Gaussian process model. We show surprising equivalences between different hierarchical Gaussian process models leading to Student-t processes, and derive a new sampling scheme for the inverse Wishart process, which helps elucidate these equivalences. Overall, we show that a Student-t process can retain the attractive properties of a Gaussian process - a nonparamet-ric representation, analytic marginal and predictive distributions, and easy model selection through covariance kernels - but has enhanced flexibility, and predictive covariances that, unlike a Gaussian process, explicitly depend on the values of training observations. We verify empirically that a Student-t process is especially useful in situations where there are changes in covariance structure, or in applications such as Bayesian optimization, where accurate predictive covariances are critical for good performance. These advantages come at no additional computational cost over Gaussian processes.