Teaching Responsibility
LJMU Schools involved in Delivery:
Computer Science and Mathematics
Learning Methods
Lecture
Practical
Module Offerings
6109STATS-JAN-CTY
Aims
To enable the student to explore the structure of multidimensional data sets.
To introduce the student to inferential procedures using multivariate data.
To enable the student to explore the structure of data in the form of a time series, and make forecasts of future observations that will arise in the time series.
Learning Outcomes
1.
Perform exploratory numerical and graphical analysis of a multivariate dataset and time series of data using appropriate statistical software.
2.
Evaluate situations in which a multivariate approach is required and conduct appropriate inferential procedures.
3.
Construct probability models for time series and use them to obtain forecasts and prediction intervals for time series data.
4.
Synthesise smoothing based methods to obtain forecasts and prediction intervals for a time series of data.
5.
Present statistical arguments effectively using a variety of media.
Module Content
Outline Syllabus:1) Multivariate Analysis
• Graphical display and numerical summary of multivariate data. Investigation of the dependence among variables.
• Discrimination and prediction. Error rate estimation.
• Hypothesis construction and testing. Use of simultaneous confidence intervals. Principal Components Analysis.
• Use appropriate software for data exploration, visualisation, parameter estimation and significance testing.
2) Time Series
• Smoothing methods.
• Moving Averages, EWMA, Exponential Smoothing, Holt’s method, the Holt-Winters method (with both additive and multiplicative seasonality).
• Box-Jenkins method.
• Identification, estimation and diagnostic checking of potential models, point and interval forecasts.
Additional Information:This final year module advances beyond univariate statistical methods to the analysis of data sets with multiple dependent variables (multivariate data).
A time series is a set of observations made sequentially through time, with the special feature that successive observations are not usually independent. The extent to which this can be both a boon and a hindrance will be discussed.