There are many computer games, learning environments, online tutoring systems or computerized tools which keeps the track of the user while learning or engaging in the activities. This paper presents results from an exploratory study and aims to group students regarding their behavior data while solving the Einstein's riddle. 45 undergraduate students were given this logic puzzle as a complex cognitive task without any time limitation. After completing the task, they were asked to report their mental effort. While grouping the similar students, cluster analysis with X-Means algorithm was used. Features such as task performance, puzzle's difficulty levels, item movements inside and between the puzzle's sections, duration and a total number of incorrect moves were used while grouping students. At the end of the lab session, six out of forty-five participants solved the puzzle and find the correct answer, on the other hand, other students reached different completion levels. Based on cluster analysis students grouped into three different clusters, Cluster_0, Cluster_1 and Cluster_2. Cluster_2 was the successful group with the highest score, lowest moves and errors, medium level of mental effort in the shortest time period. Cluster_0 had the medium level of success with highest moves and errors, the highest level of mental effort in the highest time period. Cluster_1 was the least successful group with the lowest score, medium level of moves and errors and lowest level of mental effort. [For full proceedings, see ED571332.]