Comparative Analysis of Deep Learning and Machine Learning Approaches for Air Quality Prediction

Air quality prediction is crucial for addressing environmental challenges and mitigating health risks caused by pollution. This study provides a comparative analysis of Machine Learning (ML) and Deep Learning (DL) models for air quality classification. The dataset, comprising 5,000 samples from Dhaka, Bangladesh, was used to evaluate the performance of three ML models Random Forest, Support Vector Machine (SVM), and k-Nearest Neighbors (KNN) and three DL models VGG19, VGG16, and CNN1D. Key metrics, including accuracy, precision, recall, and F1-score, were assessed to determine model performance. Results indicate that Random Forest achieved the highest accuracy among ML models (96%), demonstrating efficiency and stability in resource-constrained environments. Among DL models, VGG19 outperformed others with a comparable accuracy of 95%, showcasing its ability to handle complex patterns in air quality data. Learning curves further illustrated the rapid convergence of DL models, particularly VGG19, while ML models exhibited consistent performance across varying dataset sizes. This research highlights the trade-offs between computational efficiency and predictive performance in ML and DL approaches. It also underscores the importance of dataset diversity and scalability in building robust air quality prediction systems. Future work will explore hybrid techniques combining ML and DL to enhance adaptability and generalizability, as well as extend the dataset to include samples from multiple regions for more comprehensive analysis.