How much do we spend to save? Calculating the embodied carbon costs of retrofit

The drive to reduce carbon emissions from domestic housing has led to a recent shift of focus from new-­‐build to retrofit. However there are two significant differences. Firstly more work is needed to retrofit existing housing to the same energy efficiency standards as new-­‐build. Secondly the remaining length of service life is potentially shorter. This implies that the capital expenditure – both financial and carbon -­‐ of retrofit may be disproportionate to the savings gained over the remaining life. However the Government’s definition of low and zero carbon continues to exclude the capital (embodied) carbon costs of construction, which has resulted in a lack of data for comparison. The paper addresses this gap by reporting the embodied carbon costs of retrofitting four individual pilot properties in Rampton Drift, part of an Eco-­‐Town Demonstrator Project in Cambridgeshire. Through collecting details of the materials used and their journeys from manufacturer to site, the paper conducts a ‘cradle-­‐to-­‐gate’ life cycle carbon assessment for each property. The embodied carbon figures are calculated using a software tool being developed by the Centre for Sustainable Development at the University of Cambridge. The key aims are to assess the real embodied carbon costs of retrofit of domestic properties, and to test the new tool; it is hoped that the methodology, the tool and the specific findings will be transferable to other projects. Initial changes in operational energy as a result of the retrofit works will be reported and compared with the embodied carbon costs when presenting this paper.

Automatic construction and natural-language description of nonparametric regression models

Copyright © 2014, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. This paper presents the beginnings of an automatic statistician, focusing on regression problems. Our system explores an open-ended space of statistical models to discover a good explanation of a data set, and then produces a detailed report with figures and natural- language text. Our approach treats unknown regression functions non- parametrically using Gaussian processes, which has two important consequences. First, Gaussian processes can model functions in terms of high-level properties (e.g. smoothness, trends, periodicity, changepoints). Taken together with the compositional structure of our language of models this allows us to automatically describe functions in simple terms. Second, the use of flexible nonparametric models and a rich language for composing them in an open-ended manner also results in state- of-the-art extrapolation performance evaluated over 13 real time series data sets from various domains.