Abstract: An advertisement system that provides subscribers with the most relevant advertisements based on their individual purchase prediction of the various products and contextual information such as location, ongoing public events, festivals and personal events. The system customizes the advertisements to suit the audiovisual, language, and textual caption preferences of the subscribers. System provides for tracking of various promotional offers made to the subscribers from the time the offer is made to the time it is redeemed by the subscriber. System consists of methods to learn (a) purchase predictions (b) preferred streaming schedule and (c) the subscriber's preferences in terms of audiovisual aspects of the advertisements. Purchase prediction for individual subscriber is derived from historical purchase data by time series analysis and the influence of events such as co-purchases of other products and organized events such as sales exhibitions and festivals is derived based on Bayesian network models.