What is it about?
This article describes a project conducted in conjunction with the Central Statistics Office of Ireland in response to a planned national rollout of smart electricity metering in Ireland. We investigate how this new data source might be used for the purpose of official statistics production. This study specifically looks at the question of determining household composition from electricity smart meter data using both Neural Networks (a supervised machine learning approach) and Elastic Net Logistic regression. An overview of both classification techniques is given. Results for both approaches are presented with analysis. We find that the smart meter data alone is limited in its capability to distinguish between household categories but that it does provide some useful insights.
Featured Image
Why is it important?
Businesses and government agencies are trying to understand the challenges and opportunities to extract insights from new data sources. Our paper highlights some of the opportunities and challenges in using consumers' electricity usage data.
Perspectives
While we estimated the number of people in a household using smart meter data, this work gave us pause for thought on how such sources might be used in the future and caused us to reflect on user privacy concerns weighed up against governmental objectives. We also had to reflect on which techniques are most suitable, and how understanding of individual consumers' electricity usage patterns might help in developing climate action programmes.
Paula Carroll
University College Dublin, Ireland
Read the Original
This page is a summary of: Household Classification Using Smart Meter Data, March 2018, De Gruyter,
DOI: 10.1515/jos-2018-0001.
You can read the full text:
Contributors
The following have contributed to this page
