Realistic signal coverage loss-centric modeling and predictive analysis are key means of boosting upcoming cellular networks planning and optimizing existing ones. This work presents the results of our studies regarding the applications of the probabilistic Generalized Regression Neural Networks (GRNN) to the predictive analysis and modelling of spatial signal power loss data collected over commercial LTE networks interface in outdoor environment. The explored GRNN model is trained with the measured spatial signal power loss data obtained in three different outdoor signal propagation environments. The results of the prediction made by the proposed model showed a greater agreement with the measurements compared to the conventional least square (LS) regression modelling approach.