Spatial machine learning in R 2400-ZEWW982

his is a very innovative course that combines machine learning and spatial analysis methods and puts them into practice in R. The course shows how to work with different types of geo-localised data: points, pixels, grids, rasters, images, regions, lines, etc. - how to integrate them into coherent databases and extract information and knowledge. The course focuses on applications in economics. The course is based on the book 'Spatial Machine Learning in R' (Routledge, 2026), published by the lecturers teaching the course.

Detailed topics for the classes:
1 Working with spatial data - R language classes, reading, visualisation,
2. working with spatial data - integration of spatial data of different types and granularity (grids, points, rasters, regions, lines)
3. working with spatial data - obtaining information from the environment (construction of spatial weighting matrices, radial zoning, k nearest neighbours)
4. work with spatial data - classification of territories by density using the DEGURBA model
5. working with spatial data - machine learning methods for spatial interpolation and handling missing data
Unsupervised spatial machine learning - clustering of geolocation points (Clark-Evans test, DBSCAN, QDC)
7. unsupervised spatial machine learning - methods for measuring spatial agglomeration (ETA, SPAG)
8 Unsupervised machine learning in space - surface modelling methods (nuclear density estimation, relative spatial risk)
9 Unsupervised machine learning in space - comparison of spatial distributions (clustered KDE, spatial series clustering)
10. unsupervised machine learning in space - spatial associative rules (rules in time and space)
11. supervised spatial machine learning - global models using ambient information (random forest, artificial neural network, spatial and radial weight matrix construction)
12. supervised spatial machine learning - local models on geographically targeted subsamples (geographically weighted random forest)
13. supervised spatial machine learning - convolutional neural networks for temporal-spatial data
14. supervised machine learning in space - causality-based machine learning models
15. kriging as a method for extrapolating the results of machine learning models from points to surfaces