DAMSL-211 Time series analysis

Syllabus

The course presents predictive models for time series. The first part f the course presents Time Series Regression whereas the second part focuses on ARIMA
models. R software (the lingua franca of Statistics) is utilized throughout the course, in a wide variety of examples, based on real and simulated data.

• Linear Regression with One Predictor; Inference in Regression and Correlation
• Diagnostics and Remedial Measures; Simultaneous Inferences
• Matrix Approach to Linear Regression; Multiple Regression 
• Models for Quantitative and Qualitative Predictors; Model Selection and Validation; Diagnostics
• Weighted Least Squares; Serial Correlation
• Exploratory Analysis of Time Series Data; Time Series Regression 
• Stationarity
• ARMA models
• ARMA model Identification and Maximum Likelihood Estimation
• Forecasting with ARMA models
• ARIMA and seasonal models
• Multivariate Time Series
• Forecasting Experiments; Rolling-Window Estimation and Evaluation

Learning Outcomes

• Upon successful completion of this course students will develop an in-depth understanding of linear regression and ARIMA models.
• Students will have learned alternative ARIMA and time series regression model-building strategies, depending on the number of available predictors.
• Students will gain hands-on experience in classic (e.g. exploratory analyses, diagnostic checks) and modern topics (e.g. cross-validation, quantile regression, penalized estimation) by analyzing real (or simulated) datasets using statistical software.