Modern Applied Statistics: Learning II

Modern statistical machine learning topics moving beyond linear regression and classification. Decision trees (boosting, random forests) and deep learning techniques for non-linear regression and classification tasks.

Discovering patterns and low-dimensional structure via unsupervised learning, including clustering, EM algorithm, PCA and factor analysis, (variational) autoencoding methods, and matrix factorization.

Time series and sequence modeling via state space models and deep learning methods (recurrent neural networks, seq2seq models, transformers). Students entering the course are assumed to have foundational working knowledge in statistics, probability, and basic machine learning concepts, though the course has been designed to provide a broadly accessible treatment of the topics covered.

Topics Include

• Decision trees, boosting, random forests
• Deep learning
• Clustering
• Dimensionality structure via unsupervised learning
• Time series sequence modeling