A data-driven approach for establishing a CO2 emission benchmark for a multi-family housing complex using data mining techniques

Abstract

To reduce CO2 emissions in the building sector, South Korea uses an operational rating system, an indicator for evaluating CO2 emission performance. To conduct a reasonable operational rating, it is necessary to develop a rational and reliable CO2 emission (CE) benchmark for buildings. The conventional CE benchmarks, however, have limitations accounting for regional differences of multi-family housing complexes (MFHCs). Thus, a separate CE benchmark is required for each region for improving the rationale and reliability of the conventional CE benchmarks. To solve this problem, a data-driven approach for establishing a CE benchmark using data mining techniques was applied in this study. Data on a total of 1,212 MFHCs were established, and a total of 11 CE benchmarks (central region: 7; southern region: 4) for MFHCs were established based on the decision tree. The developed CE benchmarks were then validated using statistical methods (Mann-Whitney test, Kruskal-Wallis test, etc.). Compared to the average operational rating based on conventional CE benchmarks, the average operational rating based on the newly developed CE benchmarks decreased by 1.85% in the central region, and increased by 5.19% in the southern region, respectively. This means that the unreliability and irrationality of the conventional operational rating system (ORS) can be solved by the established ORS. The established ORS, based on the newly developed CE benchmarks, can help policymakers select and manage MFHCs with poor CE performance.