Qingdao Middle School Classroom Environmental Monitoring and Window-Opening Behavior Dataset

This dataset comprises high-resolution environmental monitoring and window-opening behavior data collected from a middle school classroom in Qingdao, China (cold-climate region), over four seasons in 2025. The monitoring campaign covered 84 school days across spring (April, 22 days), summer (June, 21 days), autumn (October, 18 days), and winter (December, 23 days), yielding a total of 120,960 raw data records. The dataset includes continuous measurements of indoor and outdoor air temperature (°C), relative humidity (%), indoor CO₂ concentration (ppm), indoor and outdoor PM₂.₅ (μg/m³), indoor and outdoor PM₁₀ (μg/m³), outdoor CO₂ concentration (ppm), and window-opening status (binary: open/closed). Environmental parameters were recorded at 1-minute intervals using Xingzong Intelligent AM308-470M industrial-grade sensors for indoor monitoring and HOBO U12-012 data loggers for outdoor measurements. Window status was tracked via magnetic contact switches mounted on the classroom's exterior windows. The merged Excel file contains 10 worksheets: (1) All_Raw_Data: 56,760 rows of merged raw monitoring data from all four seasons with a 'Season' identifier column (sampling interval: 1 minute during school hours, 6:30–18:30). (2) All_Calculated_Data: 2,640 rows of aggregated statistical summaries (hourly and daily averages) with a 'Season' identifier column. (3–10) Season-specific sheets (Spring_Raw, Spring_Calculated, Summer_Raw, Summer_Calculated, Autumn_Raw, Autumn_Calculated, Winter_Raw, Winter_Calculated) preserving original file structure for reproducibility. This dataset supports machine learning-based analysis of window-opening behavior as a climate-resilient adaptive mechanism under institutionalized teaching schedules. It has been used to train a Random Forest model (AUC-ROC = 0.897, F1-score = 0.749) with SHAP (SHapley Additive exPlanations) analysis to identify threshold-driven behavioral patterns across four daily activity phases: arrival, break, class, and leaving. Variable list (All_Raw_Data sheet): - Season: Spring / Summer / Autumn / Winter - Timestamp: Date and time (YYYY-MM-DD HH:MM:SS) - Indoor_Temperature_C: Indoor air temperature (°C) - Outdoor_Temperature_C: Outdoor air temperature (°C) - Indoor_CO2_ppm: Indoor CO₂ concentration (ppm) - Outdoor_CO2_ppm: Outdoor CO₂ concentration (ppm) - Indoor_PM2_5_ugm3: Indoor PM₂.₅ concentration (μg/m³) - Outdoor_PM2_5_ugm3: Outdoor PM₂.₅ concentration (μg/m³) - Indoor_PM10_ugm3: Indoor PM₁₀ concentration (μg/m³) - Outdoor_PM10_ugm3: Outdoor PM₁₀ concentration (μg/m³) - Indoor_RH_percent: Indoor relative humidity (%) - Outdoor_RH_percent: Outdoor relative humidity (%) - Window_Status: Window opening state (1 = open, 0 = closed)本数据集采集自中国青岛（寒冷气候区域）某中学教室在2025年四季的高分辨率环境监测数据与开窗行为数据，覆盖春季（4月，22个教学日）、夏季（6月，21个教学日）、秋季（10月，18个教学日）、冬季（12月，23个教学日）共计84个教学日，总计获得120960条原始数据记录。 本数据集包含室内外空气温度（单位：℃）、相对湿度（单位：%）、室内二氧化碳（CO₂）浓度（单位：ppm）、室内外PM₂.₅（单位：μg/m³）、室内外PM₁₀（单位：μg/m³）、室外CO₂浓度（单位：ppm）以及开窗状态（二元变量：开启/关闭）的连续监测数据。环境参数的采集频率为1分钟/条，其中室内监测采用星纵智能AM308-470M工业级传感器，室外监测采用HOBO U12-012数据记录仪；开窗状态通过安装于教室外窗的磁性接触开关进行追踪记录。 合并后的Excel文件包含10个工作表： (1) All_Raw_Data：涵盖四季全部合并原始监测数据，共56760行，带有"Season"标识列；采样时段为教学时段6:30–18:30，采样间隔为1分钟。 (2) All_Calculated_Data：包含2640条聚合统计汇总数据（按小时、日均值统计），带有"Season"标识列。 (3)–(10) 分季节工作表：Spring_Raw、Spring_Calculated、Summer_Raw、Summer_Calculated、Autumn_Raw、Autumn_Calculated、Winter_Raw、Winter_Calculated，保留原始文件结构以确保研究可复现。 本数据集可支撑基于机器学习的开窗行为分析，该行为作为制度化教学日程下的气候适应性韧性机制。本数据集曾用于训练随机森林（Random Forest）模型，其AUC-ROC值为0.897、F1值为0.749，并通过SHAP (SHapley Additive exPlanations)分析识别出四个日常活动阶段（到校、休息、上课、离校）下的阈值驱动行为模式。 All_Raw_Data工作表的字段说明如下： - Season：季节（春/夏/秋/冬） - Timestamp：日期与时间（格式：YYYY-MM-DD HH:MM:SS） - Indoor_Temperature_C：室内空气温度（单位：℃） - Outdoor_Temperature_C：室外空气温度（单位：℃） - Indoor_CO2_ppm：室内CO₂浓度（单位：ppm） - Outdoor_CO2_ppm：室外CO₂浓度（单位：ppm） - Indoor_PM2_5_ugm3：室内PM₂.₅浓度（单位：μg/m³） - Outdoor_PM2_5_ugm3：室外PM₂.₅浓度（单位：μg/m³） - Indoor_PM10_ugm3：室内PM₁₀浓度（单位：μg/m³） - Outdoor_PM10_ugm3：室外PM₁₀浓度（单位：μg/m³） - Indoor_RH_percent：室内相对湿度（单位：%） - Outdoor_RH_percent：室外相对湿度（单位：%） - Window_Status：开窗状态（1=开启，0=关闭）