Modeling large financial databases covering different assets, sectors, and regions, using linear and nonlinear
methods for dimensionality reduction, clustering, and low-rank matrix estimation. Effect of introducing parsimony
and variable selection, from PCA to generative models. Supervised and unsupervised learning techniques will be
applied to financial data. Emphasis will be placed on operational metrics, classic L2 distances, and others such as
Wasserstein distance. All of these compression techniques will be put into perspective in terms of practical
applications in investment.
