<p>Small and medium-sized manufacturing facilities often rely on limited utility data for energy management, creating a need for reliable and interpretable modeling approaches that can support practical decision-making. This study evaluates the application of Lean Energy Analysis (LEA), a changepoint regression method, for modeling weather-dependent electricity and natural gas consumption using monthly utility data. A controlled comparison was conducted between LEA and a Random Forest model using identical input variables to reflect realistic data availability conditions. Results from a chemical manufacturing case study show that LEA achieved strong predictive performance (R² = 0.874 for electricity and 0.981 for natural gas) and stable error characteristics under low-dimensional inputs. In addition to predictive capability, LEA provides physically interpretable parameters, including baseline load, changepoint temperature, and weather sensitivity, which can be directly used to estimate potential energy savings. The framework was applied to evaluate operational improvements related to fume hood management, identifying potential annual reductions of 191,259 kWh, 729 MMBtu, and 129 tons of CO₂. These findings demonstrate that transparent, low-data modeling approaches can provide effective baseline development and retrofit decision support, offering a practical pathway for implementing energy analytics in data-constrained manufacturing environments.</p>