Welcome to the Stream Mining and Urban Data Systems Group
Motivation
More and more data are available as data streams and reveal constant changes in a growing number of
phenomena. This raises the question of how to gain new insight by analysing such data and how to
automate data analysis with machine learning for data streams.
This can be exemplified by data documenting the evolution of urban transport systems.
Such continuous evolution has an impact on city services and is reflected, e.g. in the changing
level of service of transport modes. Ideally, the transformation of urban environments and systems
is expected to gradually contribute to more sustainable cities, e.g., by stimulating citizens to
choose more environmentally friendly travel modes.
Our works
We work on all aspects of developing machine learning methods prepared under demanding real settings, such as delayed access to ground truth data used to validate past predictions or the limited number of ground truth records illustrating human decisions. To develop and validate our solutions, we develop complex software platforms. These platforms are used to collect, preprocess and analyse large volumes of heterogeneous urban data, such as mobility data. This includes the use of relational databases, “big data” storage technologies, data streaming technologies, transportation network analysis services, and data analytics engines. We collect and process a full spectrum of data, including tabular data, data from social surveys, sensor data, GPS traces, schedules, and spatial data of the built environment. We combine and fuse these data to provide new insight into the evolution of urban areas.
Maciej Grzenda, PhD, DSc is the leader of the group. The group unites the works of researchers, PhD and MSc students in Data Science, software developers and software architects with backgrounds in inter alia computer science, machine learning, statistics, geodesy and geographical information systems.
Maciej Grzenda
Associate Professor, PhD, DSc
Leader of the group
maciej.grzenda@pw.edu.pl
Vision
The vision of the Stream Mining and Urban Data Systems Group is to combine research on machine learning methods for data streams with building software platforms, systems, and applications, combining storage and analytical platforms for urban data processing. By applying novel stream mining methods, we aim to address the dynamics of urban systems. To address challenges faced by inter alia urban environments, we develop novel data preprocessing and machine learning methods. By applying Data Science methods to evolving urban systems, we aim to contribute to the social good.
