The objective of this study is to develop a data-driven prediction model and sensor calibration method in variable air volume (VAV) terminal unit systems. Based on the operational data of VAV terminal unit systems, indoor loads and carbon dioxide (CO2) concentrations were predicted, and a Bayesian markov chain monte carlo (MCMC)-based sensor calibration method was used. The operational data of a VAV terminal unit used for data-driven model analysis and development were collected using the dynamic energy simulation tool, TRNSYS 17. Data analysis and Bayesian MCMC algorithms were analyzed and developed using R Studio. The indoor comfort and energy consumption were analyzed for the offset error effect of supply air flow rate in VAV terminal units. A sensor error distance function was developed using a prediction model. The sensor calibration method was developed using the Bayesian MCMC algorithm. The performance evaluation of sensor calibration methods utilized simulations data. It is confirmed that sensor calibration was confirmed to be possible in case of supply air flow rate sensor errors.