Pre-processing for missing data: A hybrid approach to air pollution prediction in Macau

Abstract

Recently, as an important issue in both urban and industrial areas due to the rapid development in economics, more and more conceptions in air pollution have been studied, and consequently forecasting the air pollution index (API) becomes increasingly important. In the past decades, researchers proposed various methods to predict the API based on previous observed data. On the other hand, however, missing of the observed data always occurs in practice and it may deteriorate the prediction performance. How to handle the missing data is often a challenge in API forecasting. This paper presents a method for pre-processing the missing observed data by adopting the multiple imputation technique for Macau API prediction using the Adaptive Neuro-Fuzzy Inference System (ANFIS). The forecasting performance after missing data pre-processing is compared with the conventional case without pre-processing and the results in terms of the root mean square error (RMSE) shows effectiveness in API forecasting against nine-years measured data in the Macau City.