Computational Modeling

Catalog description

This course focuses on programming skills for modeling data and on computationally intensive statistical methods. Topics covered include gradient descent and other optimization techniques, nonlinear least squares estimation, method of moments, and various bootstrapping and Monte Carlo techniques, including Markov Chain Monte Carlo/Hamiltonian Monte Carlo (MCMC/HMC) for Bayesian data analysis. Emphasis is placed on building probability models for data and using computational techniques to solve for their parameters. The course primarily uses the R programming language, along with the JAGS and Stan engines for MCMC and HMC, respectively, though we may occasionally “get our hands dirty” with Python, Java, or C++, as needed. By the end of the course, students should be reasonably comfortable with programming algorithms for various flavours of likelihood-based data analysis.