New Calendar-SPCA method learns interpretable electricity consumption patterns

Researchers have developed Calendar-SPCA, a novel method for learning interpretable representations of multi-periodic electricity consumption data. This technique incorporates known daily, weekly, and annual cycles directly into the learning process by structuring the feature domain as a Cartesian product of cyclic calendar axes. Calendar-SPCA utilizes an L1 loading penalty and graph total variation to achieve sparse and locally coherent loading patterns that are directly readable in their original temporal coordinates. Evaluations on GoiEner and Low Carbon London smart-meter datasets demonstrate that Calendar-SPCA effectively preserves explained variance while producing highly sparse components, outperforming classical sparse PCA methods by organizing latent factors into coherent calendar patterns. AI

IMPACT Introduces a novel method for analyzing time-series data with periodic structures, potentially improving energy consumption forecasting and management.

RANK_REASON Academic paper detailing a new method for representation learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Calendar-SPCA
• Carlos Quesada Granja
• CORE Recommender
• GoiEner
• Hugging Face
• Low Carbon London
• principal component analysis
• SPCA-TV

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →