|M.Sc Student||Blum Shem Tov Ilil|
|Subject||Investigation of Travel Habits Using Cell-Phones Data|
|Department||Department of Civil and Environmental Engineering||Supervisor||Professor Shlomo Bekhor|
|Full Thesis text - in Hebrew|
The limitations of customary methods of data collection for transportation issues and the opportunity of widespread cellular phone technology triggered this research. Literature indicates a great potential of using cellular phone as future transportation sensors. This thesis investigates selected travel patterns using passive information from cellular phones.
The data used for the investigation includes records of about 10,000 cellular-phone users at one work week. The data source was provided by a major cellular phone company in Israel. The tracking system was based on recording events that contain a change in the position of the cellular phone with respect to a given antenna. The data collected contained the following information: (a) unique ID for a given telephone number, (b) coordinates of the antenna that is serving the cellular phone, and (c) time stamp. Since the raw data collected is not designed for transportation purposes a comprehensive data processing was developed and included statistic data processing and sophisticated SQL queries to extract the transportation information out of the cell-phone data.
There are a few limitations related to the specific data set used. For example, the frequency of the observations is not always sufficient to record short trips. In addition, there are two phenomena that are specifically related to the cellular phone technology:
1. There are sometimes “zigzag” patterns that do not represent a movement but the phone is being recorded at several antennas because of network load in an antenna or other technical reasons which result in handover although there is actually no movement.
2. The phone location is recorded up to the cell-id or antenna’s’ location so the information of the location of the cell-phone itself is not accurate.
The extraction of the transportation information form the cellular phone data is not trivial. The massive amount of data requires working in a database environment. Since the dataset does not contain any information about the cellular phone user, several assumptions are needed to locate anchor points, such as place of residence and commuter location.
After calculating home and commuter locations, several trip patterns were able to be computed, such as trip rates, trip length distribution, and travel statistics. The results were compared to the previous household survey and cellular phone results from the national model. The travel patterns were chosen due to the data limitations. The majority of them were found to correlate with the previous results that they were compared to.