Detailed knowledge of passenger context is essential for developing intelligent transportation systems. For example, automated ticket sales and personal routing require more information about used means of transportation than traditional time tables can offer. Here, the authors contribute to this topic by using measurements from smartphone sensors to predict (i) whether a person is inside a bus, (ii) if the person is travelling in a diesel or an electric bus, and (iii) how the person is rating the quality of the bus ride. All three tasks are worked out by using a selection of machine learning (ML) algorithms. In tandem with sensor data, collecting a digital passenger survey was conducted to add passengers' own evaluation of the quality of their bus ride. The tests showed that the context of a passenger can be predicted relatively well. However, the prediction of passenger satisfaction is a complex task that requires further research. This research aims to give a good premise for such efforts.