DataSpark's papers accepted to 1st ICDM workshop on Data-driven Intelligent Transportation

Papers that we submitted earlier this year were recently accepted to the first IEEE International Conference on Data Mining (ICDM) workshop on Data-driven Intelligent Transportation. The conference is one of the leading research conferences on data mining and will be held in Singapore this year. The workshop seeks to explore how large-scale city data can be used in developing a more intelligent transport system. It has sought interesting papers that examine techniques in utilising this data and in mining data required to improve our transport systems.

The papers that we have submitted and have been accepted to the workshop are:

A comparative study of urban mobility pa‚tterns using large-scale spatio-temporal data, By The Anh Dang, Jodi Chiam, and Ying Li, accepted to Workshop on Data-driven Intelligent Transportation, IEEE International Conference on Data Mining (ICDM), 2018

The large scale spatio-temporal data brought about by the ubiquitous wireless networks, mobile phones, and GPS devices present a fertile ground for studying human mobility. These data sources come with high coverage and resolution that enable studies of mobility patterns for human populations at large that other conventional methods such as surveys are not capable of. In this paper, we study anonymized spatio- temporal data from telco networks to understand the variability in human mobility behavior across different geographical regions. We present methodologies for extracting trips and other mobility features from large-scale spatio-temporal data. We also look into daily activity patterns of the populations in two specific cities, Singapore and Sydney. Our results include measures of distance and frequency of people’s travel, as well as their purpose of travel, mode of transport, and route choice. We extract mobility patterns known as motifs. We also define a mobility index to assess the mobility level of individuals and compare it among different regions and demographic groups. This work contributes to a more comprehensive understanding of urban dynamics, supporting smart city development and sustainable urbanization.

Predicting MRT trips in Singapore by Creating a Mobility Behavior Model based on GSM Data, By Emin Aksehirli and Ying Li, accepted to Workshop on Data-driven Intelligent Transportation, IEEE International Conference on Data Mining (ICDM), 2018

Singapore has a significantly high coverage of both public transportation and communication network. Island-wide Mass Rapid Transport (MRT) system is the backbone of public transport, any improvement to its service or any disruption can create a considerable impact to business productivity and people’s lives.

In this paper, we discuss our implementation of a system that uses telecommunication (Telco) data to predict trips in Singapore’s MRT network in real-time. First, we study the predictive capabilities of available data sources and decide on the specific data to use for this work. We then investigate the predictability of MRT riders’ mobility behavior based on what can be observed from our data source. We then analyze the applicability of the features along with what is feasible to predict. Finally, the predictive capabilities of our supervised methods and their ensembles are evaluated.

The workshop will be held on 17 November 2018 in Singapore. More details can be found here. See you there!

Find out more about DataSpark's offerings here!

No items found.