Nowadays we are witnesses to the rapid expansion of smart cities and smart transportation services. Smart transportation services especially in the urban area can be different, but they have the same goal, to enhance the driver experiences in the urban area. Smart transportation services have constant growth. This growth is facilitated by the deployment of long-range wireless technologies such as LoRa and LoRaWAN. At the same time, the concepts of Edge computing shape the architecture and services of future smart city applications. This paper describes the implementation methodology of machine learning on the ESP-32 NodeMCU edge devices. The proposed methodology includes the implementation of classifiers such as Random Forrest, Decision Tree, XGBoost, Gradient Boost, and SVC with the scikit-learn library and Python. Scikit-learn-based models are pre-trained on PC and further processed with a micromlgen library to build an Arduino code for implementation on ESP32-based edge devices. The findings of this paper can be implemented in any smart transportation service where the prediction of wireless signal coverage for mobile users is important.