Model Failure or Data Corruption? Exploring Inconsistencies in Building Energy Ratings with Self-Supervised Contrastive Learning
Qian Xiao, Dan Liu, Kevin Credit
TL;DR
This work tackles the reliability of Building Energy Rating (BER) assessments, which are susceptible to missing values and faulty measurements. It introduces CLEAR, a self-supervised contrastive-learning framework based on SCARF that learns latent representations from building-measurement data without requiring BER labels, followed by latent-space analysis via PCA to detect rating inconsistencies and potential data corruption. Key findings include evidence of inconsistent BER assessments and data corruption in a large Irish EPC dataset, with latent-space structures and neighboring representations revealing mismatches across adjacent BER levels. The approach offers a path toward more transparent and trustworthy BER processes, enabling better data quality control and more reliable energy-efficiency policy targeting.
Abstract
Building Energy Rating (BER) stands as a pivotal metric, enabling building owners, policymakers, and urban planners to understand the energy-saving potential through improving building energy efficiency. As such, enhancing buildings' BER levels is expected to directly contribute to the reduction of carbon emissions and promote climate improvement. Nonetheless, the BER assessment process is vulnerable to missing and inaccurate measurements. In this study, we introduce \texttt{CLEAR}, a data-driven approach designed to scrutinize the inconsistencies in BER assessments through self-supervised contrastive learning. We validated the effectiveness of \texttt{CLEAR} using a dataset representing Irish building stocks. Our experiments uncovered evidence of inconsistent BER assessments, highlighting measurement data corruption within this real-world dataset.