Described are a system, method, and computer program product for generating real-time predictions for vehicle transportation requests using machine learning. The method includes generating, with a processor and a machine-learning classification model, a transportation categorization for each consumer of a plurality of consumers based on historic transaction data; processing a plurality of new transactions by each consumer, each new transaction associated with a geographic node of activity; in response to processing the new transactions, generating a plurality of vehicle transportation predictions for the consumers based on the transportation categorization for each consumer and a geographic node of activity associated with a new transaction, each vehicle transportation prediction representing a likelihood that the consumer will request vehicle transportation subsequent to conducting the new transaction; and generating a supply map interface comprising a visual identification of a location in which a number of requests for vehicle transportation is predicted to increase based on the vehicle transportation predictions.