An air quality index (AQI) is calculated based on other atmospheric pollutants and interprets how polluted the air currently is. With increasing air pollution, implementing efficient air quality monitoring models that collect information about air pollution concentration and assess air pollution in each area is necessary. Prediction methods for air quality forecasting have attracted much attention due to their performance and flexible capability. In this paper, an ensemble prediction methodology called the geographically weighted predictor method (GWP) combines the best version of machine learning algorithms and uses the additional predictor variables for prediction at the hourly level. The proposed method is employed on the Bangkok Air Quality dataset. Compared with regular machine learning models, the proposed method has better prediction performance in all prediction horizons. The results obtained suggest that such a novel model may help to enhance the accuracy of AQI prediction.

空气质量指数 (AQI) 是根据其他大气污染物计算得出的，并解释了当前空气的污染程度。随着空气污染的加剧，有必要实施有效的空气质量监测模型，收集空气污染浓度信息并评估每个区域的空气污染。空气质量预报方法因其性能和灵活性而备受关注。在本文中，一种称为地理加权预测器方法 (GWP) 的集成预测方法结合了机器学习算法的最佳版本，并使用额外的预测变量进行每小时级别的预测。所提出的方法用于曼谷空气质量数据集。与常规机器学习模型相比，所提出的方法在所有预测范围内都有更好的预测性能。获得的结果表明，这种新颖的模型可能有助于提高 AQI 预测的准确性。