Abstract
Air quality in public transport is a critical factor directly affecting urban health, yet for Almaty this issue has remained understudied until now. In this study, a systematic monitoring of CO₂, PM₂.₅, PM₁₀, temperature, and relative humidity concentrations was conducted in buses across three key transport corridors of the city. Data collection was carried out using a mobile IoT device of our own design, «Tynys», calibrated against a commercial reference sensor to ensure high measurement reliability. Special attention was paid to comparing peak and off-peak hours, which allowed us to identify patterns of air quality changes under varying passenger loads. The analysis showed that during morning peak hours, CO₂ and fine particulate matter concentrations significantly exceeded off-peak levels, confirming the key role of passenger flow in shaping in-vehicle air quality. Additionally, by applying machine learning methods, it was established that CO₂ and PM₂.₅ are the most informative indicators of pollution, enabling reliable classification of high-load periods. The obtained results fill an existing research gap in studies of air quality in public transport in Central Asia and provide a basis for further work aimed at modernizing ventilation systems, optimizing passenger flows, and implementing intelligent microclimate monitoring technologies in buses.