Constructing dynamic residential energy lifestyles using Latent Dirichlet Allocation

The rapid expansion of Advanced Meter Infrastructure (AMI) has dramatically altered the energy information landscape. However, our ability to use this generate actionable insights about residential electricity demand remains limited. In research, we propose and test a new framework for understanding by using dynamic lifestyles approach that is iterative highly extensible. To obtain lifestyles, develop novel applies Latent Dirichlet Allocation (LDA), method commonly used inferring latent topical structure text data, extract series household attributes. By doing so, provide perspective on consumption where each characterized mixture attributes form building blocks identifying sparse collection lifestyles. We examine running experiments one year hourly smart meter data from 60,000 households six describe general daily patterns. then clustering techniques derive distinct lifestyle profiles attribute proportions. Our also flexible varying time interval lengths, seasonally (Autumn, Winter, Spring, Summer) track dynamics within across find around 73% manifest multiple year. These are compared different characteristics, discuss their practical applications response program design change analysis.