Fundaments of a data stream mining platform for road quality evaluation using smartphone sensors

This paper describes a preliminary feasibility study on creating a service oriented web-smartphone platform for stream data mining. We briefly address the device orientation corrections, which are the most common problem in acquiring accurate and valid data from the sensing application. Some implementations are already described for a partial task, the road quality analysis in real-time, as a proof of concept. Thus, the main task of this paper is to state, whether it is possible to get required data for later road quality recognition and classification. The implemented android application currently stores and shares sensor data of a single agent (smartphone), in order to determine the quality, complexity and description of a road segment. First samples of measured data are analyzed with data mining software RapidMiner, in order to provide an answer to the hypothesis.