Teaching

Course name: Introduction to statistical methods for spatial data analysis

Learning objectives
Spatial data is ubiquitous in our society. In recent years, spatial data has grown rapidly in diversity and size with the swift advancement in sensing techniques (e.g. instruments onboard satellites, drones, self-driving cars) and citizen science (e.g. data from twitter, OpenStreetMaps). Core to many Geoscientific applications are statistical analysis tasks such as spatial prediction, image classification, and change detection. This course is a dive into details of the statistical methods for spatial analysis in real-life applications. During the course, students will develop strong statistical thinking and learn to develop their own Geoscientific projects to solve real-life problems. Students are expected to independently design a scientific workflow in face of a problem in Geoscience and can answer questions such as: is a statistical analysis necessary? Which dataset is to be obtained? How to describe the data? Which statistical model may be useful? What are the prospective limitations of the data and the models? What prediction accuracy can the model reach with the data available? How to interpret the model?

Learning content
After an introduction to spatiotemporal data analysis, we will focus on several fundamental statistics and machine learning concepts in regression and classification problems. The course concludes with an overview of the frontiers in spatial analysis and we will see that a large portion of state-of-the-art Geoscientific data science methods are rooted in the methods introduced in this course. The programming language Python will be introduced and exercise will be given. With the exercises, the students will learn basic Python programming and how to use it for spatial data exploration and analysis in both simulated and real-life applications.

Schedule (Download)