Video consumption behaviour is going through a paradigm shift with the viewer freed from both the confines of the living room and the limits of linear TV. While there have been advances in the user interface for over-the-top content, interfaces for broadcast television are still antiquated. This is best exemplified by the traditional grid-style Electronic Program Guide (EPG) even as the number of listed channels and associated programs has grown significantly. Technology developments in the area of consumer devices as well as network infrastructure have opened the room for significant innovation in the area of rich navigation interfaces. We believe that a personalised user interface which actively responds to viewing behaviour is within reach and towards this goal, we describe a machine learning based framework. Simulation results with selected set of attributes that describe viewer behaviour suggest that the approach is promising.