Cluster analysis or clustering is a set of techniques in the toolbox of unsupervised learning. The aim of cluster analysis is to categorize the input data, to segment it into homogeneous sub groups that guarantee a simultaneous heterogeneity across the sound subgroups. Cluster analysis is often done in an experimental data analysis framework where one likes to discover structure, e.g. on similar data objects, and where one likes to interpret observations and sub groups based on their relative distances. Cluster analysis may lead in a further step to a discriminant analysis based on labels attached to found subgroups. This course on „all about clustering“ covers several methods on clustering: these include k-means, k-expecttile clustering but also spectral clustering and minimum spanning trees.
This clustering course has several courselets covering spectral clustering, hierarchical clustering, k-expectile clustering and several applications like „understanding jumps in high frequency digital asset markets“. It also introduces into the „financial risk meter for emerging markets“ and covers „dynamic crypto networks“
Enjoy this course!
Raul Cristian Bâg and is a PhD student at HU Berlin. His general research interests are Machine Learning, MLOps, Research Transparency and Reproducibility and Decentralised Finance. He currently has the role of Teaching Assistant at the Faculty of Business Administration in foreign languages where he will be teaching seminars that include a range of subjects from Statistics and Data Analytics to Blockchain. Currently, he also works part time as a senior cloud engineer at a private corporate company. This helps him keep in touch with the latest industry trends and technologies.