The advancement in the field of transportation has been accelerating with policymakers shifting the focus towards autonomous and connected vehicles. Being an important component of the IoT infrastructure, a majority of its deployment is still based on the results of conventional approaches and measurement practice. Factors and the best conditions are identified using trial-and-error approaches which are not entirely accurate. Therefore, a thorough understanding of the influences that affect the performance of the windshield tag and developing a prediction model based on the most influential factor has been presented in this paper. In addition, two machine learning algorithms are proposed to achieve the objective of the paper. Linear regression analysis provides a reliable estimation of the RSSI levels of the windshield tag from all the selected location. The second proposed algorithm is based on the logistic regression model which provides an accurate prediction model based on tag readability by the reader. The predicted tag detectability model coupled with the RSSI level estimation model gives a complete overview of the behavior of the RFID tag when deployed on the windshield of the vehicle. This will serve as a guideline for the RFID transportation application which may extensively rely on RFID as a part of the IoT transformation.