The short-term power load prediction of single households is a challenging issue in the research fields of Smart Grid (SG) management/planning, viable energy usage, energy saving and the bidding system design of electricity market. The reason for this is the unpredictability and uncertainty in electricity consumption pattern of individual household. The energy management/planning of the SGs is even becoming more complex due to the integration of Distributed Energy Resources (DERs). The DERs are useful in decreasing the bill of the electricity consumer by empowering them to produce their own green energy. With the huge development in Advanced Metering Infrastructure (AMI), Big Data (BD) and machine learning models, the potential benefits of dynamic pricing schemes and DERs can be fully accomplished. But, the accurate prediction of power generated through DERs as well as forecasting the user power profile is a big issue. The user power profile varies hourly, daily, weekly and seasonally due to the various environmental and seasonal effects. In this work, the focus is on exploring and evaluating machine learning models for accurately predicting user power profile for energy management in a smart community. Eight regression models are evaluated for the prediction of the power consumption of the single household. The simulation results indicate that the Radial Basis Function (RBF) kernel is the most suitable machine learning model for forecasting the short term power consumption of the single household.