Occupant behaviors influence the energy consumption of dwelling mechanical ventilation systems significantly. There is still a lack of effective method to analyze the occupant behaviors in adjusting the mechanical ventilation systems in buildings. Therefore, this study proposes a data mining-based method to reveal the occupant behavior patterns and the motivations behind. A first derivative Gaussian filter-based approach is developed to detect when an occupant increases or decreases the mechanical ventilation flowrate without direct measurements. A logistic regression-based statistical analysis approach is developed to find the crucial factors influencing the behaviors of increasing and decreasing ventilation flowrate. A K-means clustering-based analysis approach is introduced to further find the motivations behind the behaviors. The proposed data mining-based method discovers the ventilation behaviors and the crucial factors influencing them successfully for the occupants from the 10 dwellings located in a Dutch community. The motivation patterns of the ventilation flowrate adjustment behaviors are further revealed based on the discovered crucial factors. The discovered insights are useful to provide more accurate assumptions and inputs for the mechanical ventilation system models. It is also helpful to generate tailored design, refurbishment and control strategies.