Methods and apparatus for predicting communications resources which will be needed at a venue and then controlling the amount of available resources dynamically are described. In various embodiments real time or near real time video of areas of the venue are used to predict the number of people in a portion of a venue and/or the direction of movement. Along with other information such as the type of event and/or event schedule collected information is supplied to a set of trained resource requirement models which are used to predict future resource needs at a venue, e.g., while an event is ongoing. Commands are sent to dynamically vary the amount of communications resources provided to one or more portions of the venue. Resources which can be varied included but are not limited to fixed wired WAN bandwidth, WiFi bandwidth, cellular bandwidth, network based on-demand services, transcoding services, firewall services, etc.