Buildings are becoming taller,increasing complexity in indoor environmental conditions and occupantperceptions. This study classifies floor level in a multi‑storeyeducational building in Dhaka, Bangladesh, using environmental measurements andoccupant perception data. A total of 1,087 observations were collected overfour months during the summer via physical surveys and air‑qualitymonitors. Three machine learning classifiers such as Decision Tree, RandomForest, and XGBoost were trained and evaluated using confusion matrices; SHAP(Shapley Additive exPlanations) was used for feature ranking and modelinterpretability. All models achieved accuracies above 96%. The mostinfluential features for floor‑level classification were numberof fans, number of lights, humidity, temperature, and CO2 concentration. Thesefindings demonstrate the capability of interpretable ML to distinguish verticalvariations in indoor environments and identify key drivers of inter‑floordifferences. The results offer actionable insights to inform the design ofadaptive, occupant‑centric, and energy‑efficient multi‑storeyeducational buildings.