This paper proposes a methodology to identify passages of a rail vehicle on switches and crossings using GNSS measurement and the railway infrastructure database of the considered region. Given the imprecision of GNSS measurements, a map matching task is performed. Then, a database containing all identified switch passages is populated, containing construction parameters of the switch as well as passage parameters such as the route taken or the speed of the vehicle. Further, the switch passages are clustered in representative categories. For this, a dimension reduction using the t-SNE algorithm is performed and an agglomerative clustering algorithm is used. The methodology is applied to 2400 km of measured rides of a German regional train, resulting in 2139 turnout passages, classified in 22 clusters.