Technologies are described herein for determining characteristics of a plurality of people at an event. A method may include receiving location information about a plurality of active mobile phones, receiving sentiment information relating to the plurality of people, the sentiment information including at least one identifier of the event, determining at least one characteristic about the plurality of people based at least in part on the location information and the sentiment information, aggregating the sentiment information, classifying the aggregated sentiment information as either favorable or unfavorable with respect to the event, and calculating a statistical confidence level indicating an extent the aggregated sentiment information of the particular group correlates with the plurality of people. The location information may include at least one identifier of a geographical location and a time period. The location information and sentiment information may be collected from a particular group of the plurality of people.