Abstract: To analyze the driving style of drivers at the on-ramp junction of the expressway, this paper employs the NGSIM dataset. To ensure the accuracy of data, spatiotemporal constraints are set to eliminate abnormal data and the smoothed trajectory data in on-ramp merging area are obtained. First, factor analysis (FA) method is employed to reduce the dimensionality of the original multidimensional features and five major factors well represented in the driving style are obtained. Second, the K-means algorithm is used to cluster the principal factors and three driving styles are identified: cautious, prudent and aggressive driving. The recognition results before and after FA are compared. Our results show in aggressive driving, continuous lane-change within a short period of time and the smaller headway during the whole ramp merging process are more likely to occur.