In consequence of varied topology, geographic scale of about 3.28 million sq. km and wide varieties of climate, India is prone to frequent floods which impact millions of lives every year. Efficient flood prediction system is the need of the hour to better prepare for future disasters. A lot of research has been conducted in this domain, where solutions majorly revolved around Geographic Information Systems (GIS) and remote sensing-based mapping, geospatial frequency ratio, multi-criteria decision-making, statistical indexes and some applications of machine learning and deep learning as well. But all of these solutions have lacked geospatial validation, and scalability and also have neglected the complexities, class imbalances and patterns present in the data. Thus, this research discusses Geospatial Flood Prediction, specifically focusing on the Brahmaputra region incorporating NetCDF data of geographical features and historical flood-related data. The model proposed focuses on data handling (normalization, handling outliers), data conversion (NetCDF to CSV) to ensure scalability and reliability and uses advanced machine learning algorithms like Multi-Layer Perceptron (MLP). This study helps not only in predictions but also in understanding the impact of various geographical factors like digital elevation, topographical wetness index (TWI), precipitation, etc. on the probability of flooding. The findings of the study tend to be in a positive direction with the model reaching an accuracy of almost 94% and suggesting that Filled Digital Elevation affects the probability of floods the most (correlation of 0.608). This model in the future could serve as a helping hand to disaster management agencies in predicting areas that are more prone to flooding so precautions can be taken timely to prevent any significant losses.