This research presents the design and implementation of an automated weather forecasting system powered by IoT and advanced machine learning models. The framework integrates environmental sensors for real-time data collection with predictive models such as ARIMA, Random Forest, and Deep Learning. By combining exploratory data analysis with resilient forecasting techniques, the system delivers highly accurate predictions. Experimental results show that Random Forest achieved near-perfect accuracy with R² = 0.9999 and MSE = 0, outperforming Deep Learning (R² = 0.9999, MSE = 0.0281) and ARIMA (R² = 0.0126, MSE = 331.9). These findings highlight Random Forest as the most effective model, emphasizing the potential of IoT-driven AI systems in building climate-resilient smart applications.

This research presents a deep learning–based framework for accurate detection of plant diseases, with a focus on mosaic and curl viruses affecting crops like tomatoes. Leveraging the VGG19 convolutional neural network through transfer learning, the model achieves 97.66% test accuracy, supported by strong evaluation metrics including precision (0.877), F1-score (0.807), and low MSE (0.018). The study uses the PlantVillage dataset, applying preprocessing and augmentation techniques to enhance robustness. Results show that VGG19 effectively identifies symptoms such as mottling, leaf curl, and chlorosis, offering reliable diagnostic support. The work contributes to sustainable agriculture by improving early disease detection, with future directions including ensemble methods, mobile deployment for farmers, and integration with multi-model frameworks.