Teaching

Machine Learning

Data Analysis

Time Series Analysis

Probability

Statistics

Python

since 2020

Machine Learning

- Advanced Data Analysis Concepts with Python

Workshop at summer school Informatica Feminale, University Freiburg / University Furtwangen, Germany

A challenging task depicts the analysis of complex data sets for the purpose of unravelling deeper insights from and predicting future patterns based on available data. Machine learning algorithms provide powerful approaches to complete the data analysis spectrum beyond the classical statistical modelling concepts. The workshop aims to enable the participants to understand and apply supervised and unsupervised learning methods and to address advanced data analysis tasks in a sustainable pythonic manner.

The focus of the workshop is the introduction of machine learning approaches used for advanced data analysis via hands-on examples covering:

• introduction to supervised and unsupervised machine learning, such as regression, classification and dimensional reduction
• discussion of statistical principles guiding machine learning concepts
• discussion of interpretability approaches to machine learning algorithms
• application of open source libraries for advanced data analysis and machine learning

since 2019

Data Analysis Concepts with Python

Workshop at summer school Informatica Feminale, University Freiburg / University Furtwangen, Germany

The basic work flow of data analysis encloses the potential pre-processing, suitable storage, beneficial visualization and statistical analysis of complex data. The goal of this workshop aims to enable the participants to design and implement computational approaches to data analysis tasks in a sustainable pythonic manner.

The main focus of the workshop lies on the concepts of data science and analysis via hands-on examples covering:

• efficient strategies for computation with and organisation of large data sets
• introduction to explorative, descriptive and inferential data analysis
• discussion of statistical principles guiding data analysis concepts
• application of open source libraries for data analysis methods
• visualization techniques of complex data sets

2021

Uncertainty in Climate Data Across Scales

Module 3 (7-11 June 2021) at the interdisciplinary online summer school Trends, Rythms and Events in the Earth's Climate System - Past, Present and Future (23 May–13 June 2021 & 22 August–12 September 2021) organized by the University of Potsdam, Germany and funded by the Volkswagen Foundation.

Particularly in the context of climate sciences, the ability of researchers to make transparent and explainable statements in the presence of uncertainty is of great importance. This module aims to provide the fundamentals of uncertainty and to foster a deeper understanding about how uncertainty impacts the analysis of climate data. The module starts with an overview of the related concepts of uncertainty, probability and risk. Then, the sources of uncertainty and the uncertainty quantification in the reconstruction process from proxy measurements to past climate indicators are discussed. Here, explorative, descriptive and inferential data analysis approaches are explained and compared, such as frequentist and Bayesian approaches. In the second part, aspects of uncertainty are discussed in the context of predictive data analysis applied on recent climate observations. The last part of the module introduces the principles of uncertainty and sensitivity analysis to investigate simulated data, such as future climate predictions.

More details can be found on the summer school homepage and at Martin Trauth's homepage.

2018

Earth Surface Dynamics - Understanding Processes at the Earth's Vulnerable Skin

Summer school (25 May–10 June 2018 & 12 August–2 September 2018) at University of Potsdam, Germany

The summer school is founded by the Volkswagen Foundation and consists of two consecutive summer school sessions for 25 doctoral students from geosciences, environmental sciences and related fields such as biology, chemistry and physics.

The summer schools will be designed for doctoral students, aiming

• to improve their skills to understand the complex interaction of the processes shaping the Earth's surface at different temporal and spatial scales,
• to monitor, model and predict the results of these interactions, and
• to identify and mitigate risks of natural and human-caused interference in these processes in an interdisciplinary and intercultural environment.

The 1st set of modules will focus on types of signals and noise commonly encountered at the Earth's surface, and methods of acquiring, processing and analyzing data with non-destructive physical surveying methods. The 2nd set of modules will be about the examination and modeling of the process underlying the data collected at the Earth's surface.

More details can be found on the summer school homepage and at Martin Trauth's homepage.