At present, motorcycles are mainly tested on the road with real drivers. In the course of shifting the scope of testing from the road to complete vehicle test rigs, the driving task of longitudinal dynamics is also being shifted to the test rig. The claim is that the real driving task will be reproduced with comparable characteristics on the test bench. The state of the art knows the description of the temporal actuation sequence of gearshifts primarily from the passenger car sector, which usually takes place according to optimal criteria such as comfort or sportiness. Actual and sometimes non-optimal driver inputs are thus not taken into account and may well deviate from the definition of "optimal" gear changes. With our method, it should be possible to parameterize a variable model for shifting operations by feature extraction and an evaluation environment to describe varying driver characteristics in reality. The gearshift model comprises activation and deactivation curves for three actuators. To parameterize the model, the necessary data are collected in a pilot study. The data are fed into the evaluation method, with initial trends being derived through exploratory data analysis. This will be used to derive parameter sets for real driving situations. The evaluation shows that the underlying model for the description of shifting processes is suitable for describing real shifting processes sufficiently accurately with few features. After the feature analysis, there are clear trends for the actuation of the clutch and throttle grip for the use in test rig environments. In summary, our method enables a real impact on the human-machine interfaces by driver inputs as on the road, even in test bench operation. In the future, driving robots of test benches can be parameterized with the presented feature sets, thus ensuring the transferability of real driving maneuvers to test benches.