Trip Purpose Inference with Movement Trajectories, Land Use Data, and Social Media Data

• Activity Number: 102 - The Essential Role of Statistics for the Future of Mobility
• Type: Invited
• Date/Time: Monday, July 31, 2017 : 8:30 AM to 10:20 AM
• Sponsor: Transportation Statistics Interest Group
• Author(s): Qing He* and Jing Gao and Yu Cui and Chuishi Meng
• Companies: University at Buffalo, SUNY and University at Buffalo, SUNY and University at Buffalo, SUNY and University at Buffalo, SUNY
• Keywords: Trip purpose identification ; Social media data ; GPS data ; Land use data

Abstract

This paper investigates the credibility and effectiveness of social media in labeling the trip purpose for the trip ends. We first identify trip ends from GPS trajectories produced in the household survey data in the Bay Area of California. In the meanwhile, we collect 4-year geo-tagged tweets in the same area from 2013 to 2017. We extract the potential locations embedded in tweets by using selected preposition words: "at", "in", "on", etc. These "tweet locations" can be further categorized into location categories classified by Google Place Types, which also provides API to nearby land use data for any given GPS location. Through our classification algorithm, we show the promises of tweets in inferring the trip purposes. Fused with land use data, the tweets data proves to be a more informative resource in trip purpose inference.