This course provides a brief introduction to UMAP (Uniform Manifold Approximation and Projection), one of the recently developed VizTech (visualization techniques). The course also covers the problems of high-dimensional clustering, the methodology of UMAP, applications of UMAP to various datasets, and a comparison of UMAP with t-SNE (t-distributed stochastic neighborhood embedding).
Among all the applications, I would like to highlight the CryptoPunk case. CryptoPunks from Larva Lab is one of the 'veteran' NFT series launched in 2017 and now has a total market capitalization of over $4 billion. UMAP helps to study the properties of CryptoPunks through its flattened pixel vectors. This leads to high-dimensional data, which we represent through modern visualization techniques to detect clusters and similarities between digital assets.
Quantlet links of all the codes and outcomes are included in the course materials.