.

This project explores the application of unsupervised learning techniques to analyze a high-dimensional dataset of household energy consumption and environmental conditions. Using Principal Component Analysis (PCA) with Varimax rotation and Non-metric Multidimensional Scaling (MDS), the study reduces 25 sensor variables into five interpretable factors. The analysis focuses on identifying the primary drivers of energy usage and detecting anomalies. A Procrustes Analysis is further employed to validate the consistency between linear and non-linear dimensionality reduction approaches, providing a framework for energy usage data diagnostics.

nothing