Big Data for Mobility Tracking Knowledge Extraction in Urban Areas
As Track&Know enters the final year of research and development, we look back over 2019 at what we have achieved. One of the major objectives of the Track&Know project is to integrate online data streams, heterogeneous, contextual and archival data on one big data platform. This enables big data experts and stakeholders to advance their operational, processing and decision-making activities. Over the past year we have developed the big mobility data integrator (BMDI) which is a fully featured industrial grade solution that is able to scale out and accommodate big data from different domains, interoperating with modern data storage technologies as well as other persistence approaches and that can support all important programming languages including Python, Java, R and Scala as well as other traditional programming approaches.
The big data platform consists of data sources and data store components, connectors together with the Communication platform an underlying infrastructure and Big Data Apps such as the big data processing (BDP) toolbox, the big data analytics (BDA) toolbox, the complex event recognition (CER) toolbox and the visual analytics (VA) toolbox which are being used in the three project pilots for Fleet Management, Car Insurance and Healthcare Services.
A number of tools have been developed for the platform including Big Data processing. The data cleansing and enrichment tool is a scalable solution for online processing of streaming mobility data, which takes as input streaming GPS traces, performs cleansing and map-matching, enriches them with selected traffic data, weathers attributes indicating road conditions (e.g., wind, rain or ice) and nearby fuel stations and other points-of-interest (POIs), which will be used in the pilots related to mobility and insurance. The tool is built on top of scalable big data processing technologies, including Kafka and Spark Streaming. Eventually, it will help improve the accidents risks estimation, hot spot analysis and electric mobility analysis.
The data cleansing and enrichment pipeline developed in Track&Know for online processing and enrichment of streaming GPS traces.Big Data Analytics, crash prediction tool which can be used to assess the risk of an individual accident predicted through a combination of mobility modelling and Artificial Intelligence techniques. State-of-art machine learning models are applied based on sophisticated indicators that capture individual mobility distribution, driving behavior events, time evolution of mobility demand. This will be used in the insurance pilot and helps estimate risk scores as well as provide individual AI-supported feedbacks for users to improve their driving safety.
Track&Know has also worked on answering business questions for the Royal Papworth Hospital (RPH, UK) regarding their services for Obstructive Sleep Apnoea (OSA) patients. Due to the high risks associated with OSA patients driving RPH want to understand whether they can reduce the travel distance for their patients. Track&Know have been analysing patient attendance, no-show rates and distances travelled and so far, the results helped RPH understand the optimum geographical location for outreach clinics helping them to improve their services.
Track&Know has been promoting their research over the year attending and presenting at prestigious events across Europe and Asia and being shortlisted the BDV PPP summit 2019 “success stories awards” and receiving the EuroVA 2019 ‘’Best Paper Award’’.
2020 will be the final year of the project and the focus will be on the final development and integration of the toolboxes in the platform and the use of these tools in the three pilot studies, we will also be looking at the sustainability of the research particularly the patenting and commercialisation of our results and via open source platforms sharing much of our research with the wider community. More information: https://trackandknowproject.eu/
TRACK&KNOW ‘Speed dating workshop’ Hasselt University Nov 2019