Journal Search Engine
The Journal of The Korea Institute of Intelligent Transport Systems Vol.7 No.6 pp.21-30
사분위편차 및 관리도 모형에 의한 GPS 수집기반 구간통행속도 데이터 이상치 제거방안 연구
The Quartile Deviation and the Control Cahart Model of Improvement Confidence for Link Travel Speed from GPS Probe Data
Abstract
The travel speed collected by the prove-car equipped with the GPS has the problems, which are the data's stability and finding out the representative travel speed, by the influence of the traffic signal and etc. at the interrupted traffic. This study was conducted to develop the method of filtering the outlier data from the data collected by the prove-car. The method to remove the outlier data from the serial data which were collected by the prove-car was adapted to each of the quartile deviation statistics model and the management graphic statistics model. The rate of removing the outlier data by the quartile deviation method was 0~3.7% while the rate by the management graphic statistic methods was 03.~7.2%. Both methods show the low removal rate at the dawn time when the traffic is inactivity, on the other hand the remove rate is high during the daytime. However, both methods have the problem such that the threshold level for removing the outlier data was established at the low bound in the case as good as the statistics model. Therefore, it is required for the experience calibration.