Unraveling Power Consumption Pattern Using Multidimensional Scaling and Clustering Analysis

Abstract

Understanding regional and seasonal patterns in electricity consumption is essential for ensuring grid reliability, resiliency, and sustainable infrastructure management. With the rapid growth in electricity demand data and the integration of distributed energy resources, traditional methods struggle to capture zonal load patterns effectively. This study presents a dimensionality reduction approach combined with clustering to analyze seasonal energy load dynamics across zones managed by the New York Independent System Operator (NYISO). By applying Multidimensional Scaling (MDS), we condense complex high-dimensional data into a form that reveals nuanced relationships among zones. This transformation allows for effective K-means clustering of load profiles, uncovering four distinct seasonal clusters representing shared load characteristics among NYISO zones. Our analysis, covering eleven zones across four seasons, provides visually compelling two- and three-dimensional scatter plots that highlight both converging and diverging demand patterns. This approach unveils similarities in electricity usage and significant seasonal variations, offering deeper insights into the regional and temporal complexity of electricity demand across NYISO’s operational landscape. The Python script developed for this work is available on GitHub and the signalsciencelab.com website.

Keywords

• Energy
• NYISO
• electricity consumption pattern
• load characteristics
• multidimensional scaling
• clustering