The adoption of Building Performance Simulation for decision-making during the early phases of design would help improve the building’s lifecycle performance. One important challenge, however, is that most building parameters are not yet finalized during such early phases; hence, an extensive number of design scenarios will exist to be simulated. Given that mathematical simulation of building performance is a computationally expensive problem, the simulation of many building scenarios becomes unfeasible during the design phase. Worst of all, a considerable number of such design scenarios end up being inferior to the code baselines and are only wasting computational time and power. To decrease the number of scenarios to be simulated during the decision-making of building parameters, this paper uses machine learning to build a recommender system, capable of identifying and filtering out the scenarios with worse performance than the building code’s baseline, without performing energy simulation. This goal is achieved by training six different classifiers and selecting the one that best achieves the objective. After model evaluation and comparison, it was found that the artificial neural network (ANN) model has the best performance among the tested classifiers. The ANN model is hence used for the development of a recommender system to support simulation-based decision-making for selecting design parameters during the early phases of building design.