PM2.5 Prediction based Weather Forecast Information and Missingness Challenges: A Case Study Industrial and Metropolis Areas

Abstract

Air Quality Index (AQI) is one of the indicators used to identify risks associated with the air pollution that impact daily living. Recently, Thailand is one of the countries facing air quality problems with an increasing level of fine particulate matter (PM2.5) which has become a negative trend in the news and social media. The collected air quality information from sources are limited since they consist of missing data. Further study and research regarding missing data are required to obtain preventive and corrective actions. This paper has collected historical data from world weather online application programming interface to analyze the correlation between the various factors. The intuitive imputation algorithm called “Iterative Imputation based on Missingness Pattern Analysis (II-MPA),” using weather forecast data as a correlated factor to PM2.5 to impute the missing historical data. The comparison results show significant improvement in both imputation and prediction with the RMSE of 4.0 when using an unbiased dataset. This provides an alternative and fundamental concept for dealing with missing air quality data and is also a reachable predictive model for PM2.5 without using complex scientific data.