The linear hybrid machine learning model achieves the expected target well. The overall inversion accuracy (R²) of the model is 0.84, and the RMSE is 12.92 µg m⁻³. R² was above 0.7 in more than 70 % of the sites, whereas RMSE and mean absolute error were below 20 and 15 µg m⁻³, respectively. There was severe pollution in winter with an average PM_2.5 concentration of 62.10 µg m⁻³. However, there was only slight pollution in summer with an average PM_2.5 concentration of 47.39 µg m⁻³.