As the number of mobile subscribers of long-term evolution (LTE) service increases, it becomes important for various parties, such as operators, policymakers, and researchers, to examine how well LTE cells are deployed in terms of actual performance. To this end, we focus on spectral efficiency (SE) with the cell edge user throughput (TP) and average cell SE which can be calculated from the spectrum data of an LTE physical downlink control channel decoding device—Rohde & Schwarz TSME. For these two aspects, crucial probabilities for the performance evaluations are defined using a joint distribution of resource block utilization and cell TP. We derive novel transformation methods that make them approximately follow a joint Gaussian distribution and use it to compute the probabilities. Furthermore, a deep neural network is adopted to analyze not only limited cases but also a wider range.