Skip to main content
. 2024 Oct 30;24(21):6983. doi: 10.3390/s24216983
Algorithm 4 ML Pipeline for AQI Prediction and Feature Importance

Libraries Required: Pandas, Seaborn, Matplotlib, Scikit-learn

Input: Meteorological data from CSV files

Output: AQI predictions and feature importance

  •   1:

    Step 1: Data Loading

  •   2:
    Load CSV files into DataFrame dffinal:
    dffinal=pd.concat([df1,df2,⋯,dfn])
  •   3:

    Step 2: Data Exploration

  •   4:

    Visualize relationships with pair plot and correlation heatmap:

  •   5:

    Step 3: Model Training

  •   6:
    Define features X (e.g., visibility, temperature) and target y (AQI):
    X={features},y=AQI
  •   7:
    Train models like linear regression and extra trees regressor:
    y^=Xβ(forlinearregression)
  •   8:

    Evaluate models with metrics such as MSE, as mentioned in Equation (1)

  •   9:

    Step 4: Feature Importance

  • 10:
    Calculate and visualize feature importance:
    Ij=1ntrees∑i=1ntreesIji
  • 11:

    Plot top 10 important features