Computational tools for exploratory data analysis
Computational tools for exploratory data analysis
Prof. Dr. E. Müller, Dr. M. Sips, Dr. P. Köthur - Deutsches GeoForschungsZentrum GFZ
Friday, October 2, 2015
GFZ, Telegrafenberg, 14473 Potsdam
Buildings Haus H, Room VR3 and A27, GeoLab
Geosciences have to deal with a growing amount of data coming from various sources such as sensors, simulation models or geo-archives. Extracting meaningful information from all these data is a challenge that cannot be met by traditional analysis methods alone. Data intensive approaches such as data mining or visual data exploration are essential methods to analyze large and complex data.
The course will cover basic technologies in data mining and visual data exploration. Such methodologies in data analytics are of major interest for geoscientists in different research fields and widely applicable to industrial settings. Data ranges from satellite images, hydrological sensors, atmospheric measures, up to social media data. In all of these areas users try to analyze large datasets by extracting interesting dependencies, patterns, or unexpected events. The course will give an introduction to the knowledge discovery process, efficient clustering algorithms, open source data mining tools, as well as exploration of time series and high dimensional data.
The benefits of these methodologies will be demonstrated by several synthetic and real-world examples. Hands-on exercises using open source tools will allow students to gather experience in using data-intensive analytics methods. Furthermore, the course is open for discussions of problem setting that students have in their own research field. We envision such an exchange of ideas, challenges, and methodologies to be an ideal setting for an interdisciplinary course.
Morning session
Data-Intensive Science – a new research paradigm
Visual Data Exploration – concept, tools, and examples from geoscience
Data Mining – methods and application examples
Afternoon session
Hands-on exercises with WEKA, KNIME, and GGobi