Window opening behavior significantly impacts building energy consumption, indoor environmental quality, and thermal comfort. During summer months, occupant window operation patterns substantially affect both building energy performance and personal comfort levels. This study examines residential window-opening behavior in Dalian, China, aiming to develop a probabilistic prediction model. Through a three-month monitoring campaign across 10 residential buildings, we systematically evaluated influencing factors including thermal conditions (indoor/outdoor temperature), humidity, PM2.5 levels, and occupant habits. Key findings demonstrate that indoor temperature constitutes the predominant determinant of summer window operation behavior. We implemented two machine learning approaches – support vector machines (SVM) and logistic regression (LR) – for behavior prediction, with both models exhibiting robust performance (accuracy > 85%). Comparative analysis revealed marginally superior predictive capability in the SVM model compared to LR. This research contributes both empirical data characterizing summer window operation patterns in Dalian’s residential sector and theoretical support for energy-conscious architectural design and thermal environment management. The developed model enables more effective natural ventilation strategies and indoor climate optimization, facilitating measurable reductions in building energy demand.