Network-collected call trace data has the potential to provide relevant information that can be leveraged to enhance the accuracy of network planning and optimization algorithms.In this paper, we present a geolocation technique that relies on prediction-based RF Pattern Matching (RFPM) and its application in the generation of high-accuracy traffic hotspot maps. The method relies solely on call trace data and an uptodate configuration of the cellular network. We demonstrate the value of this algorithm over existing solutions and validate the approach with LTE measurements collected in a very dense urban environment (Tokyo center). Unlike existing techniques, the proposed solution is capable of identifying traffic hotspots at the level of individual buildings and the resulting map is consistent with building heights.