Analysis of Subway Passenger Flow for a Smarter City: Knowledge Extraction From Seoul Metro’s ‘Untraceable’ Big Data

Abstract

Timely and efficient analysis of big data collected from various gateways installed in a smart city is an intractable problem and requires immediate priority. Given the stochastic and massive nature of big data, the existing literature often relies on artificial intelligence techniques based on information theory. As a new approach, this paper presents a knowledge extraction method based on an analysis of Seoul Metro's 'untraceable' ridership big data. Without identification information, the untraceable ridership data only shows the hourly accumulation of station entry and exit information. To reconstruct the missing information in the data set, this study proposes a fluid dynamics model and adopts a heuristic genetic algorithm based on optimization theory as the problem solver. The result of our model presents the distribution of the elapsed time defined on an hourly basis taken until a passenger returns to the station they departed from. To validate our model, we acquired subway ridership data with passengers' identification with permission from Seoul Metro. This paper presents two novel aspects of subway ridership, namely the dependency on departure time and the discrepancy between weekend and weekday traffic. Our analytical approach contributes to solving the problem of extracting hidden knowledge from big collection of data missing critical information, e.g., constantly and autonomously gathered data fragments from numerous gateways in smart cities. INDEX TERMS Inverse problem, genetic algorithm (GA), optimization, wave decomposition, harmony search algorithm, mass conservation law, data mining, outdoor duration time, Seoul metro subway ridership.

Investigating spatio-temporal mobility patterns and changes in metro usage under the impact of COVID-19 using Taipei Metro smart card data

“…Metro smart card trip data has been combined with socio-economic data gained from surveys to predict the demographic attributes of smart card users (Zhang and Cheng 2018 ), to infer the employment status of passengers (Zhang and Cheng 2020 ), to analyze the associations between demographic attributes and travel patterns (Goulet-Langlois et al 2016 ), and to identify commuters among passengers of a metro line (Mei et al 2020 ). Other studies have used metro smart card trip data to predict lifestyles of passengers based on trip patterns (Shin 2020 ), to identify home and/or work locations of passengers (Li et al 2015 ; Sari Aslam et al 2019 ; Zou et al 2018 ), to assess the spatial distribution of commuting trips and jobs-housing ratios (Zheng et al 2019 ), and to infer trip purpose and activity (Lee and Hickman 2014 ; Zou et al 2018 ).…”

Automatic Identification and Quantification of Safety Risks Embedded in Design Stage: A Bim-Enhanced Approach

“…There is no advance control method applicable to the design stage, so it can no longer meet the requirements of newly subway safety management. In addition, the current risk identification process is still through 2D drawings and the design scope considered in the current design specifications are far from being able to cover all kinds of safety problems in the construction and operation phases (such as the safety of employees during construction (Xia et al, 2020;Zhou & Guo, 2020), the safety of staff and passengers during normal operation (Danfeng & Jing, 2019;Shin, 2020), the safety of public evacuation of large passenger flow under emergencies (Chen et al, 2020;Chen, 2020;Cheng et al, 2020;Li et al, 2017, etc. ), lack of visually risk identification and quantification methods, can no longer meet the increasingly complex needs of projects.…”

Analysis of college students’ canteen consumption by broad learning clustering: A case study in Guangdong Province, China

“…More specifically, the students’ canteen data simply records the consumption amount at each time the students spend. The data is fragmented and dynamically increasing, which is bond to raise some challenging issues that probably hinder the computational model optimization [ 18 ]. Then some multi-variate statistical methods should be introduced into the design of the machine learning algorithmic flow.…”