Effectively and Efficiently Supporting Predictive Big Data Analytics over Open Big Data in the Transportation Sector: A Bayesian Network Framework

Effectively and Efficiently Supporting Predictive Big Data Analytics over Open Big Data in the Transportation Sector: A Bayesian Network Framework

연구 분야: Databases

학회: 2022 IEEE Intl Conf on Dependable, Autonomic and Secure Computing, Intl Conf on Pervasive Intelligence and Computing, Intl Conf on Cloud and Big Data Computing, Intl Conf on Cyber Science and Technology Congress (DASC/PiCom/CBDCom/CyberSciTech)

Today, various types of valuable data can be collected with ease and at a rapid pace. In recent years, many governments, researchers, and organizations have been driven by open data pioneers, to make their data available for public. Transportation data, such as public bus performance data, is an example of open big data. The analyzing of these open big data can be used in social services. For example, bus service operators might get a vision into time delays in bus services by processing and mining public bus performance data. Then, making ameliorative steps (e.g., adding more buses, rerouting some bus routes, etc.) results in improving the feeling of the passenger. We provide a Bayesian framework, which is applied on big data obtained from transportation system. Specifically, a number of Bayesian networks have been used in our framework to predict whether a bus will arrive late or early at a specific bus stop. We investigate and establish the optimum network settings and/or parameter permutations for each (bus stop, bus route, arrival time)-triplet. The results demonstrate that the proposed Bayesian framework effectively supports predictive analytics on big transportation data collected from the City of Winnipeg, Manitoba, Canada.