Identification of the most influential areas for air pollution control using XGBoost and Grid Importance Rank

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journal

View graph of relations

Author(s)

  • Jun Ma
  • Jack C.P. Cheng
  • Zherui Xu
  • Keyu Chen
  • Changqing Lin

Detail(s)

Original languageEnglish
Article number122835
Journal / PublicationJournal of Cleaner Production
Volume274
Online published12 Jul 2020
Publication statusPublished - 20 Nov 2020

Abstract

Due to the rising concern about air quality, air pollution prediction and control has been a hot research domain for scholars in recent years. Many studies have been conducted to predict and control air pollution using different kinds of methods. However, these studies did not explore the air quality interactions between areas and areas. They cannot answer questions like “which area would have a more substantial spatial influence on others?“, and “which area should be of focus when controlling the air pollution considering the air movements?” To identify the most influential areas for air pollution control can effectively benefit policymaking and achieve better results. To this end, this study proposes a methodology framework combining XGBoost and Grid Importance Rank (GIR). The GIR technique is inspired by the Google page rank algorithm, which is widely used in ranking web pages based on their influences. Combined with the mechanism of the variable importance in XGBoost, the proposed method can identify the areas that have the most substantial influence on others, and these areas should be of focus when controlling the air quality. A case study in the northwestern U.S. is conduced to validate our methodology. The results show that XGBoost can well model air pollution interactions between areas and areas. The modeling R-square of PM2.5 forecasting can reach 0.9631. The importance map indicates that the government should give priority to control air pollution in southern Oregon considering the impact of this region on the northwestern U.S.

Research Area(s)

  • Air pollution control, Grid importance rank, The most influential areas, XGBoost