HCNCS: An IoT Based Logical Real-Time Weather Data Analysis to Predict Air Quality by Using Hybrid Convolutional Neural Classification Scheme

Universal air pollution creates serious global issues which destroy public well-being together with environmental conditions and life quality standards. The research introduces HCNCS which stands for Hybrid Convolutional Neural Classification Scheme while acting as an IoT-based real-time solution for weather data research to achieve accurate air quality predictions. A network of IoT sensors in focused geographic areas collected instant data about weather factors including temperature and humidity and wind velocity and atmospheric pressure together with PM2.5 and PM10 levels from public agencies and environmental bodies. A set of data processing techniques involving data cleaning then normalization and feature engineering processes were applied to improve model accuracy levels. The HCNCS architecture merges Convolutional Neural Networks (CNNs) with Gradient Boosting for features extraction and classification to create an effective combined solution. The model reached an exceptional accuracy level of 98.29% which surpassed all three traditional models including ARIMA, SVM and Random Forest. The F1-scores from category-based performance evaluation reached more than 96% in all air quality domains while achieving outstanding results for predicting both clean and severe pollution conditions. During a 30-day evaluation period the HCNCS model displayed real-time prediction efficiency which achieved 93% accuracy in its real-time forecast. The model displays high resilience which makes it ready for extensive use in environmental monitoring systems.