Breadcrumb
Time Series Analysis
Announcement
- Lectures take place in the first teaching block
- A detailed unit description
Syllabus
Course Outline
Lecture Notes
- Chapter 0: Time series analysis essential concepts
- Chapter 1: Introduction
- Chapter 2: Simple descriptive techniques
- Chapter 3: Probability models for time series
- Chapter 4: Estimation in the time domain. An example of model identification.
- Chapter 5: Forecasting. An example of model estimation and forecast.
- Chapter 6: Frequency-based methods for time series
- Chapter 7: Spectral analysis
- Chapter 8: Bivariate processes
Practicals
Problem Sheets and Solutions
- Collaboration and Collusion Where coursework or assignments count toward the final assessment mark for a unit, collaboration with other students is not allowed (except where specifically permitted in the unit or project documentation) and is treated as a form of plagiarism called collusion. While discussing preliminary work with other students is perfectly acceptable, all the detailed working and any final write-up and results (model description, explanations, interpretations, conclusions, mathematical arguments, computer code, numerical results, etc.) must be the student’s own independent work.
- Penalties for Late Submission of Assessed Coursework For work submitted up to 24 hours after the agreed submission deadline, a penalty of ten marks out of 100 from the mark the student would have received applies (e.g. coursework that is marked at 77% would then become 67% once the penalty is applied). For work submitted after the 24 hour period but within 7 days of the agreed submission deadline, the penalty will again be 10 marks per day. For work submitted 7 or more calendar days after the submission deadline, the work will receive a mark of zero.
- Feedback on problem sheets is available on Balckboard under Course Information.
- Problem sheets: 1, 2, 3, 4, 5, 6, 7
- Solutions: 1, 2, 3, 4, 5, 6, 7
Case Studies
- Thanks to Prof. Guy Nason. This is a case study of how time series analysis may help to understand wind speed data [PDF], [R program].
- EEG, Spectral analysis and Filters
R Program and Data sets
- R: A self-learn tutorial from Prof. Peter Green
- Airline data set, monthly totals of international airline passengers (in thousands) for 1949-1960. Plot the time series.
- Recife, average air temperature (deg C) in successive months for 1953-1962. R program to plot the time series, estimate seasonal effect, and acf.
- Slutsky-Yule effect, illustrated by applying a filter to a random series.
- Johnson Johnson quarterly earning per share data set. There are 84 quarters measured from the first quarter of 1960 to the last quarter of 1980.
- Sample autocorrelation coefficients, calculated based on beaver body temperature data set.
- Fit a time series model to Lynx data set
- Fit a time series model to Wool data set
Online Resources
- Variance-stabilizing transformations, thanks to Larry Winner, Department of Statistics, University of Florida!