Name
Abstract | ||
|---|---|---|
Recent researches have put a lot of effort on mining vehicle sensing data, in order to help improve the driving safety. However, due to the lack of continuously collected data, most work assumed that one particular driver would only show or hold one driving style on any trips he/she was going on. In this paper, we analyzed more than 3.5 million GPS data points and about 68,500 driving events obtained from onboard devices installed in cars running in big cities for nine months. Different than other methods, we establish the driving style vector for each driver based on the fact that the same driver performs different behaviors on different trips. We propose a two-step clustering algorithm and obtain four representative driver groups with different driving styles. In addition to "normal", "aggressive" and "clam", we also find out an "experienced" driving style that was never mentioned in other existing works. |
Year | DOI | Venue |
|---|---|---|
2019 | 10.1007/978-3-030-29516-5_37 | INTELLIGENT SYSTEMS AND APPLICATIONS, VOL 1 |
Keywords | DocType | Volume |
Vehicle sensing data, Two-step clustering algorithm, Driving style analysis | Conference | 1037 |
ISSN | Citations | PageRank |
2194-5357 | 1 | 0.37 |
References | Authors | |
0 | 4 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Linna Wu | 1 | 1 | 0.37 |
| Huan Li | 2 | 1 | 0.37 |
| Hengtian Ding | 3 | 1 | 0.37 |
| Lizhuo Zhang | 4 | 2 | 1.43 |