A method collects data and subjects it to statistical analysis to detect localized events, which assists in network provisioning. Illustratively, the data employed is hourly network traffic counts that are collected at cell sites. By taking advantage of the additive property of a Poisson process, the method integrates spatial neighbor information by aggregating temporal data in various areas, and iteratively estimating the event location and the radius of event impact by examining the posterior probability based on the aggregated data.