With 30% of the world's final energy consumption and 40% of carbon dioxide emissions being attributed to the building sector, accurately predicting building energy consumption has become essential for various energy management applications such as identification of energy efficiency measures. With the increasing availability of smart utility meters and data related to building energy consumption, multiple techniques in Artificial Intelligence (AI) such as machine learning are being successfully applied to identify building energy usage patterns and develop focused energy efficiency plans. The UC Davis campus buildings are highly information-intensive and effective interpretation of this building data can help identify energy demand patterns, predict future energy demands, and plan varied types of energy saving strategies. This thesis performs energy demand prediction using multiple machine learning models and analyzes the prediction outcome to identify major areas to focus energy efficiency efforts at a student dining facility on campus: Segundo Dining Commons. Six machine learning models for each of the three energy commodities: Steam, Chilled Water, and Electricity were created and the best-performing model for each commodity was selected to generate a prediction of future demand. The model developed was also fed with four different meteorological scenarios: 0.5°C, 1°C, 2°C rise in outside air temperatures and a typical year with max recorded temperatures for each day of each hour in the past 5 years. The major observations made were simultaneous heating and cooling demands in summer, high energy demands even during school breaks, constant electricity demands and minor changes in demand for different temperature scenarios. Broad energy saving areas were then identified from the observations that can help develop focused energy efficiency plans.