Floods in India are caused by a mix of heavy rainfall, soil conditions, and human activities like dam operations and land-use changes. In this study, we analyzed flood events across 242 river catchments in India to understand what causes floods in different regions. We grouped the catchments based on climate and human influence, then used a machine learning model to predict flood peaks. This model also helped us identify which factors, like soil moisture or rainfall, play the biggest role in each region. Our findings show that different areas have different flood behaviors: in wetter, natural areas, soil wetness before a storm is most important, while in drier, human-modified regions, rainfall intensity and human changes matter more. This approach helps improve flood prediction and planning, especially in regions with limited data, and can support better decision-making as climate and land use continue to change.