文章基本信息

标题：MODEL FOR ESTIMATING BUS ARRIVAL TIMES BY COMPARING VARIOUS CLASSIFICATIONS
本地全文：下载
作者：HAEKAL MOZZIA PUTRA ; MOHAMMAD FAIDZUL NASRUDIN
期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2019
卷号：97
期号：21
页码：3018-3030
出版社：Journal of Theoretical and Applied
摘要：The availability of reliable and precise bus information such as bus arrival times renders public transportation more attractive. It helps passengers plan by reducing their waiting times. This paper aims to develop an estimated time of arrival (ETA) model that is based on a comparison of various classifications and groupings of real-time bus tracking data. The empirical analysis results demonstrate that the prediction accuracy differs across methods, even using the same dataset. The methodology consists of three stages: literature review and identification of existing problems; development of an ETA model; and testing and comparison of models. Data are obtained from a mobile bus tracking application, namely, BasKita, in Universiti Kebangsaan Malaysia (UKM). Data such as the route ID, bus stop ID, distance, day, time interval and log time are used as features. Groupings of data are suggested, such as daily data, data by the path and the complete data set. In this paper, linear regression (LR), artificial neural network (ANN) and sequential minimum optimization regression (SMOreg) are used to develop the model. The performances of ETA models are compared via the correlation coefficient (CC) method and in terms of the root mean square error (RMSE) and the mean absolute error (MAE). This work uses moving average (MA) technical analysis on the data to reduce the estimation error. The results that are obtained using the ANN method with daily grouping and using MA as a feature are the most accurate. The results of this study contribute to the development of an ETA model that can achieve satisfactory accuracy to increase the quality of the bus service.
关键词：Estimated Time of Arrival; Machine Learning; UKM