Enhanced Fuzzy Inference System for $P M_{10}$ Concentration Prediction Using Genetic Algorithms

This research introduces an enhanced Fuzzy Inference System (FIS) by integrating Genetic Algorithms (GA) for optimizing the rule set. The proposed GA+FIS model is applied to forecast <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\boldsymbol{P}\boldsymbol{M}_{\boldsymbol{1}\boldsymbol{0}}$</tex> concentrations in the Metropolitan Area of the Valley of Mexico (ZMVM). To validate its effectiveness, the GA+FIS model is compared with two robust machine learning models, XGBoost and LightGBM, regarding their performance in <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\boldsymbol{P}\boldsymbol{M}_{\boldsymbol{1}\boldsymbol{0}}$</tex> concentration prediction. The results demonstrate that the GA+FIS model surpassed both XGBoost and LightGBM models, achieving an RMSE of 3.500632, R2 of 0.888926, and MAE of 2.759147. Notably, the GA+FIS model exhibited the closest alignment between predicted and actual <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\boldsymbol{P}\boldsymbol{M}_{\boldsymbol{1}\boldsymbol{0}}$</tex> concentrations, with an ideal slope of 0.97, indicating superior accuracy compared to LightGBM (slope = 0.80) and XGBoost (slope = 0.75). Across sampled instances, the GA+FIS model consistently aligned well with observed data points, underscoring its robustness in capturing complex relationships within the dataset.