Overall Idea : Using machine learning and neural networks to predict floods using rainfall and waterlevels and their direction of propagation to mitigate risk by providing a webapp to officials to provide us real time data for managing the situation when no present data would be available for process.
Back-end Solution : We will be taking the past data for waterlevels and rainfall for predicting the approaching flood and then taking inputs from volunteers (wherever possible) for rainfall levels and amount of water in various regions to determine threat levels. For machine learning part we would be using microsoft cognitive toolkit and C++. We will use heat maps using pyplot for signifying the threat level.
Front-end Solution : Making a Web-app which will display three heatmaps. 1st one would be indicating the threat levels in the flood effected area's. 2nd one would help the volunteers to direct the people to much safer place which will be decided according to water clogging levels and contour height. 3rd one will predict the regions probable for floods on a larger scale. Web-app will also be taking data from volunteers (wherever possible) regarding the water levels , ration demands and the emergency resources needed for rescue operations.
Implementation : Backend using C++ and microsoft cognitive toolkit. Frontend using HTML,CSS,Javascript,Bootstrap and Django Diploying on microsoft Azure.