Increased urban population and road traffic has resulted in traffic congestion. Thus, traffic state estimation is the process of determining the current state of traffic, such as traffic volume, speed, and congestion levels, using sensor data and mathematical models. Probe vehicles equipped with Global Navigation Satellite System (GNSS) receivers, and other sensors are now frequently used to collect data on traffic condition such as speed, vehicle to vehicle spacing etc. The data collected by probe vehicles can be used to detect traffic congestion and other traffic-related issues in real-time. The probe vehicles function as the moving traffic detectors, which are not restricted to selected or fixed locations. With advent of new algorithms, the probe vehicle data can be utilized to predict traffic congestion using machine learning algorithm. In this presented research work the authors have implemented the Artificial Isolation Forest algorithm to predict the congested locations using instantaneous vehicle to vehicle space headways derived from ultrasonic sensors and speed derived from the positional data acquired using GPS sensor on the road segments subjected to traffic congestion. From the analysis of the results, it can be posited that Isolation Forest algorithm can be suitably used to detect the traffic congestion using speed and instanteneous space headway data.