In this emerging world, most of the peoples are wasting their precious time in travelling. The drastic increase in the number of vehicles is due to an increase in population. In this contribution, we propose an ontology-based moving weighted average model to forecast traffic congestion. Real-time traffic data was collected from the camera and with the help of an Arduino board using Ultrasonic and PIR sensors. The proposed system uses a moving weighted average model to calculate the threshold distance and speed from the recent history of traffic data like vehicle count, the distance between the front and back vehicles, and vehicle speed. The system interprets the semantic relationship of the collected data with the threshold data to predict traffic congestion. Due to the scalability of the data, we conduct experiments in Apache Spark. Thus, it will increase the speed of traffic prediction. The system will insist on the traffic status as a location-based message to assist the people in planning their commute.