Regression and Time Series

Catalog description

The first part of the course covers Generalized Linear Models (GLMs). Topics include exponential family, important link functions, estimations (maximum likelihood estimation, generalized moment matching), diagnostic tests for model validations (graphical methods, chi-square statistics, t and F tests, AIC and BIC, likelihood ratio test), applications of GLMs on real data, prediction and confidence intervals. It also includes penalized regression (ridge and lasso regression, k-nearest neighbors algorithm). The second part of the course covers Time Series Analysis. Topics include an introduction to discrete stochastic processes, random walks, stationary processes, autocorrelation functions, and partial autocorrelation functions, various time series models (exponential smoothing, autoregressive (AR) model, moving average (MA) model, ARMA model), autoregressive conditional heteroskedastic (ARCH) model, generalized ARCH (GARCH) model, variants of GARCH, predictions and their confidence intervals using time series models.