In this study, an attempt is made to assess the feasibility of several machine learning techniques for forecasting the energetic performance of a hybrid renewable energy unit consisting of an earth-air heat exchanger (EAX) and a building-integrated photovoltaic thermal (BPV/T) unit. The unit provides preheating/precooling of outdoor air in cold/warm days and generates electricity throughout the year. The employed methods are artificial neural network (ANN), support vector machine network (SVMN), and fuzzy network (FN). These techniques are employed to develop a relationship between the input and output parameters of the unit. The annual total energy output of the unit is taken as the essential output of the unit, while the input parameters were the length, depth, and width of the BPV/T unit, the air mass flowrate, and length and diameter of the EAX unit. The results indicated that all the methods are successful at the prediction of the annual total energy output of the unit; however, the SVMN outperforms other methods in the test phases where the non-trained data sets are examined. Finally, it is demonstrated that the SVMN model can successfully predict the output for any arbitrary combination of the inputs within the training intervals.