Extreme Weather events pose significant challenges to infra-structure, ecosystems, and communities. To address the challenges, we require advanced predictive analytics to leverage the vast data generated by modern sensor networks and climate monitoring systems. This paper introduces research focused on applying predictive analytics to improve the fore-casting capabilities of extreme weather events. The proposed research aims to use predictive analytics to analyze weather satellite data and employ an Attention-based Hybrid LSTM-FCN architecture, a fusion of Long Short-Term memory and Fully Convolutional networks to identify patterns, correlations, and potential indicators to contribute to a more accurate and timely prediction of extreme weather events. The key objectives of this paper are: (1) Develop accurate and timely predictive models for extreme weather events using big data analytics and machine learning. (2) Improve lead time for warnings and enhance impact assessments to better prepare and respond to varying weather conditions. The methodology involves data preprocessing, feature selection, and the implementation of machine learning algorithms, including but not limited to neural networks, decision trees, and ensemble methods. The research emphasizes the importance of real-time data integration and continuous model refinement to evolving climatic conditions.