An algorithm for initial public transport network design over geospatial data

Collecting geospatial data from different sources e.g. mobile phones and devices brings new opportunity to extract real needs of people in an urban ecosystem. Having data about people's everyday movements, we can understand people preferences and needs in the urban transport system. A modified transport network (or even a bunch of alternatives) can be suggested as the results of analysis. This new solution reflects needs of people and reduces transfer time and increases satisfaction level. However, the problem of geospatial data analysis is needed to be solved so that the authorities could choose (sub)optimal routes. Choosing an optimal routes network is an iterative procedure which requires human (expert) intervention. To avoid costs at the initial stage, we suggest an algorithm which helps to build initial sets of routes based on the big set of geospatial data in respect with reducing an average length cost function. Some use cases on synthetic data explain the efficiency of the algorithm over big geospatial data processing.