Verschoben - Spatial Data Science

Die jüngste Entwicklung quantitativer Methoden zur intelligenten Datenreduktion und geeigneten Analysen ist ein zentrales Thema in den Umwelt- und Sozialwissenschaften. In beiden Bereichen sind georeferenzierte numerische Daten heute massiv verfügbar und können weiter ausgebaut werden.

The main objective of SDSW2020 is to initiate a dialogue about the different issues we face when performing and developing innovative methodologies of mining, analysis, modelling and visualization of geo-spatial data.

The recent development of quantitative methods allowing to perform intelligent data reduction and suitable analysis is a central issue in environmental and socio-economic sciences. In both these fields, geo-referenced numerical data are nowadays massively available, and can be further enhanced by other sources of information numerically transformed, but this information is often complex and sometimes unstructured or noisy. Thus, discovering interesting spatial or intrinsic patterns is a challenging task that led scientists to search for new tools. With this in mind, innovative techniques based on clustering, pattern recognition and data mining, can be employed to extract knowledge and insights from data. In addition, new formalisms need to be developed to directly incorporate other sources of information in the characterization of the geographical space such as textual contents.

These theoretical developments proved to be very helpful in various applications. Here just some examples: natural hazard susceptibility assessment (e.g. flood, landslides, earthquakes, wildfires); multivariate time series analyses, for both environmental risks (e.g. pollution time series) and renewable energy potential assessments (e.g. meteorological data, such as wind speed, rainfall, solar radiation); understanding network flows (such as commuter traffic). Nevertheless, several issues need to be addressed to improve these approaches, such as, for example, information bias and information noise, scale and mapping unit, selection of the predictor variables (redundancy/irrelevance and overfitting related problems), uncertainty.

The main theoretical topics of the workshop include, but are not limited, to three principal axis:

  • Spatial quantitative methods using a strong statistical/mathematical framework, especially focused on the quality of formalism of the method;
  • Geovisualization, with a major accent on visual analytics and computational data mining techniques focused on high-dimensional attributes;
  • Pattern recognition and modelling with special emphasis on approaches based on data-mining and machine learning;

The main applications will be closely related to the research in environmental sciences, quantitative geography and spatial statistics, in particular:

  • Natural and anthropogenic hazards (e.g. flood, landslides, earthquakes, wildfires, soil, water and air pollution);
  • Renewable energy resources (wind and solar);
  • Socio-economic sciences, characterized by the spatial dimension of the data (e.g. census data, transport, commuter traffic).

Early registration: 15. März – 10. April 2020
Late registration: 10. April – 30. April 2020






Universität Lausanne, Geopolis, UniL Mouline, 1015 Lausanne

 ‭(Ausgeblendet)‬ Anmeldeformular ‭[2]‬

 ‭(Ausgeblendet)‬ Anmeldeformular ‭[1]‬

Weitere Informationen zu diesem Inhalt