Clustering daily AMI data using the two-step clustering method

Abstract

Analyzing advanced metering infrastructure (AMI) data is vital for consumers to reduce electric costs and for companies to plan their business models. Thus, the statistical analysis of AMI data has become popular. In this paper, we propose a two-step clustering method that considers the usage amount and AMI pattern. Compared to the existing clustering methods, the proposed method provides different clustering groups that can be further used for time-of-use rates in Korea. We also demonstrate the differences between each group based on the electricity price before and after tariff reform.