<p>This study aims to assess the effectiveness of various Machine learning based models in predicting the Air Quality Index (AQI) for Gurugram City, India. The Air Quality Index (AQI) pertains to many Sustainable Development Goals (SDGs), mainly those associated with health, sustainable urban development, and climate action. This analysis utilize tree-based techniques, including M5P, Random Forest, Random Tree, and Reduced Error Pruned Tree. The data set was collected from the Central Pollution Control Board (CPCB), containing daily AQI observations from January 2020 to December 2023. The first 3&#xa0;years of daily data were used for model development, and the rest of 1&#xa0;year was used for model validation. 15 different models were developed using various input combinations by the lag of the days from 1 to 15. The detailed analysis demonstrates that the M5P-based model exhibits superior predictive accuracy compared to other models in testing stage, achieving a maximum coefficient of correlation (CC) value of 0.8287. The mean absolute error (MAE) is also recorded at 33.6908, while the root mean square error (RMSE) is 45.1823. The scattering index (SI) during the testing phase is 0.2375. Results also show that M5P models outperform with only 5 input variables <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="704_2025_5535_Article_IEq1.gif" Format="GIF" Height="19" Rendition="HTML" Resolution="72" Type="Linedraw" Width="302" /> </InlineMediaObject> <EquationSource Format="TEX">\({\text{AQI}}_{\text{T}-1},{\text{AQI}}_{\text{T}-2},{\text{AQI}}_{\text{T}-3},{\text{AQI}}_{\text{T}-4},{\text{AQI}}_{\text{T}-5}\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <msub> <mtext>AQI</mtext> <mrow> <mtext>T</mtext> <mo>-</mo> <mn>1</mn> </mrow> </msub> <mo>,</mo> <msub> <mtext>AQI</mtext> <mrow> <mtext>T</mtext> <mo>-</mo> <mn>2</mn> </mrow> </msub> <mo>,</mo> <msub> <mtext>AQI</mtext> <mrow> <mtext>T</mtext> <mo>-</mo> <mn>3</mn> </mrow> </msub> <mo>,</mo> <msub> <mtext>AQI</mtext> <mrow> <mtext>T</mtext> <mo>-</mo> <mn>4</mn> </mrow> </msub> <mo>,</mo> <msub> <mtext>AQI</mtext> <mrow> <mtext>T</mtext> <mo>-</mo> <mn>5</mn> </mrow> </msub> </mrow> </math></EquationSource> </InlineEquation>. Results of the Taylor diagram show that M5P-T5 outperforms the other models, followed by RF, REP Tree, and RT. The M5P T-5 model performs better than all other models using different input combinations. In contrast, using this dataset, the RT model demonstrates the least efficacy among those utilized for predicting AQI in Gurugram City, India.&#xa0;</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Prediction of Air Quality Index (AQI) using machine learning-based tree models—a case study of Gurugram City, India

  • Akshay Kumar,
  • Parveen Sihag,
  • Karan Singh,
  • Fatemeh Esmaeilbeiki,
  • Balraj Singh,
  • Hari Om

摘要

This study aims to assess the effectiveness of various Machine learning based models in predicting the Air Quality Index (AQI) for Gurugram City, India. The Air Quality Index (AQI) pertains to many Sustainable Development Goals (SDGs), mainly those associated with health, sustainable urban development, and climate action. This analysis utilize tree-based techniques, including M5P, Random Forest, Random Tree, and Reduced Error Pruned Tree. The data set was collected from the Central Pollution Control Board (CPCB), containing daily AQI observations from January 2020 to December 2023. The first 3 years of daily data were used for model development, and the rest of 1 year was used for model validation. 15 different models were developed using various input combinations by the lag of the days from 1 to 15. The detailed analysis demonstrates that the M5P-based model exhibits superior predictive accuracy compared to other models in testing stage, achieving a maximum coefficient of correlation (CC) value of 0.8287. The mean absolute error (MAE) is also recorded at 33.6908, while the root mean square error (RMSE) is 45.1823. The scattering index (SI) during the testing phase is 0.2375. Results also show that M5P models outperform with only 5 input variables \({\text{AQI}}_{\text{T}-1},{\text{AQI}}_{\text{T}-2},{\text{AQI}}_{\text{T}-3},{\text{AQI}}_{\text{T}-4},{\text{AQI}}_{\text{T}-5}\) AQI T - 1 , AQI T - 2 , AQI T - 3 , AQI T - 4 , AQI T - 5 . Results of the Taylor diagram show that M5P-T5 outperforms the other models, followed by RF, REP Tree, and RT. The M5P T-5 model performs better than all other models using different input combinations. In contrast, using this dataset, the RT model demonstrates the least efficacy among those utilized for predicting AQI in Gurugram City, India.