Occupant presence and behavior in buildings have significant impact on space heating, cooling and ventilation demand, energy consumption of lighting and appliances, and building controls. For this reason, there is a growing interest on modeling occupant behavior, especially occupancy information. An occupancy prediction model based on an indirect approach using indoor environmental data is important due to privacy concerns and inaccurate measurements associated with the direct approach using cameras and motion sensors. However, such an indirect-approach-based occupancy prediction model has not yet fully discussed in building simulation domain. To tackle these issues, this study aims to develop an indoor environmental data-driven model for occupancy prediction using machine learning techniques.The experiments in the Building Integrated Control Test-bed (BICT) at Dankook University was conducted to collect the ground truth occupancy profiles, indoor and outdoor CO2 concentrations and electricity consumptions of lighting systems and appliances for a data mining study. The results show that the proposed indoor environmental data-driven models for occupancy prediction using the decision tree and hidden Markov model (HMM) algorithms are well suited to account for occupancy detection at the current state and occupancy prediction at the future state, respectively.