Spatio-temporal prediction and factor identification of urban air quality using support vector machine

Accurate air quality prediction can provide better supervision and reference for management policies. Due to difficulties in data acquisition, combined spatio-temporal prediction is still inconclusive. This study utilizes the support vector machine (SVM) method to predict air quality of unknown space and time. Extracted from a geographic information system (GIS), geographic features such as population, land use, economy, pollution sources, and terrain parameters were added to a time series. Temporal prediction was first executed in the reference stations, and the predicted air quality index (AQI) was then used to spatially infer the future AQI of unknown locations. Verification indicated high accuracy for short-term temporal prediction. Various meteorological and climatic effects were observed to be influential in seasonal difference. In the spatial inference stage, urbanization and city types were spatial features that appeared to impact air quality. Agriculture and forest use, transportation use, residential use, and economic factors were clearly correlated to AQIs, whereas population and labor force were not. This study establishes a prediction framework in northern Taipei based on SVM. Other locations can build their own models based on local actual data to achieve better decision-making, urban planning, or other applications. ? 2021 Elsevier B.V.