Jump to content

Large-Scale Mobility Data Analytics

Large-Scale Mobility Data Analytics

Ongoing digitalization generates a new wealth of large-scale mobility data. This provides unprecedented opportunities for analyzing and predicting mobility information, but also poses new challenges which have to be overcome: the spatio-temporal dependencies in movement data require new, highly specialized methods in order to allow for efficient processing, analysis and storage.

Established solutions like Apache Hadoop excel at handling large volumes of data, and spatial data support is provided by extensions such as GeoSpark or GeoMesa. However, movement data combine temporal and spatial aspects and are not well supported by existing tools. Implementations of such complex analyses are being developed in the data science community, but they are usually confined to comparatively small datasets. We fill this gap and integrate advanced spatio-temporal data analytics techniques into distributed data management and processing landscapes, thus offering a solution for efficiently processing large-scale movement data.

 

 

We provide a one-stop shop for efficient knowledge extraction from large-scale spatiotemporal mobility & traffic data.

 

 


Explorative Data Analysis


Machine Learning and Algorithm Development


Professional Software and System Engineering

Ship movements shown on a map

Explorative Data Analysis

We assess data quality, investigate the potential for knowledge extraction, and find and handle data gaps and inconsistencies.

We explore the potential of your mobility datasets and unveil possible pitfalls prior to advanced data-driven analysis.

Transaction data by data volume and complexity

Machine Learning and Algorithm Development

We apply classical machine learning and deep learning techniques and develop novel customized algorithms for movement data. To ensure quality and performance, we thoroughly validate the knowledge extraction algorithms and identify solutions for distributed computing.

We efficiently capture and exploit the structure inherent in your mobility datasets with scientifically and technically proven machine learning approaches for knowledge extraction.

Transaction data drawn on a map of Vienna

Professional Software and System Engineering

We develop and integrate software modules. For best performance, we optimize algorithm runtimes and design distributed IT infrastructures (including cloud computing).

Our software for knowledge extraction is well-designed and implemented and may be readily integrated with your existing IT infrastructure.

 

References

ASFINAG logo

 

 

Real-time travel time prediction for the Austrian highway system

> 4300 km of highways, 915 road sections, 955 traffic detectors (VMIS), 950 travel time sensors

Explorative Data Analysis:

Integration of traffic detectors, travel time sensors, traffic event database (incidents, congestions, road construction) and weather reports

Machine Learning and Algorithm Development:

Real-time traffic state prediction based on traffic sensor data

- Short-term prediction (next 4 hours)

- Long-term prediction (up to 2 weeks)

Professional Software and System Engineering:

Integration of the developed Python-based system into the client’s production environment

Magenta business logo

 

 

 

Movement pattern extraction from anonymized mobile phone data sets

Billions of records per day

Explorative Data Analysis:

Data quality assessment

Investigation of movement patterns

Machine Learning and Algorithm Development:

Trajectory-based CDR (call detail records) analysis algorithms, such as :

  • Travel demand analysis (including origin-destination matrices, travel route and transportation mode inference, traffic flow volumes and passenger counts
  • Functional land use classification

Professional Software and System Engineering:

Integration into the client’s existing computing framework (Apache Hadoop cluster)

Runtime optimization of Spark code

Frequentis logo

 

 

Movement pattern extraction from anonymized mobile phone data sets

Billions of records per day

Explorative Data Analysis:

Data quality assessment

Investigation of movement patterns

Machine Learning and Algorithm Development:

Trajectory-based CDR (call detail records) analysis algorithms, such as :

  • Travel demand analysis (including origin-destination matrices, travel route and transportation mode inference, traffic flow volumes and passenger counts
  • Functional land use classification

Professional Software and System Engineering:

Integration into the client’s existing computing framework (Apache Hadoop cluster)

Runtime optimization of Spark code

 

Research Field