Engineering
ST500: Advanced Engineering Statistics (5 ECTS)
This module will provide a second level coverage of statistics with an emphasis on topics of use to engineers and practical hands-on experience of applied statistics using statistical software.
Taught in Semester(s) I. Examined in Semester(s) I.
Workload: 40 hours (24 Lecture hours, 6 Tutorial hours, 10 Lab hours).
Module Learning Outcomes.
On successful completion of this module the learner should be able to:
- Use appropriate graphical and numerical summaries for different data types;
- Formulate a simple linear regression model, make statistical inferences on the fitted model using both parameter estimates and the ANOVA table;
- Calculate regression diagnostics and use these to check model assumptions, including linearity, normality, constant variance, independence and the presence of outliers and influential points;
- Formulate a multiple regression model, interpret output from a regression procedure and use this to test hypotheses on the overall model, individual terms, and make model comparisons;
- Incorporate categorical variables in regression models using factors/indicator variables and interpret and test associated parameters;
- Evaluate the effectiveness of simple experimental designs; Assess the importance of different sources of variability in terms of components of variation;
- Use a response surface model to determine optimal settings and product improvement;
- Compare times to failure (breakdown) for two groups with censored data;
- Apply control charts to monitor a production process.
Indicative Content
Main topics include:
Revision of basic probability models (binomial, Poisson, normal/Gaussian) and principles of applied statistics (estimates, confidence intervals, hypothesis tests); Concepts of association and causation; simple linear regression; General linear model: continuous and categorical explanatory variables, estimation, tests of fit and individual terms, ANOVA table, diagnostics, model checking, transformations, model comparison and model selection; Principles of statistical experimental design: randomisation, replication, blocking, factorial experiments, fractional factorials; Analysis of standard simple experiments, ANOVA, multiple comparisons, analysis of covariance; Response surfaces, estimation, interaction; Reliability, simple models (exponential, Weibull, extreme value), censoring mechanisms Quality inspection and control, simple quality charts, design for improvement;Open channel flow
Lab Sessions:
Computer laboratory sessions will be a key component of this module. These will allow hands-on experience of applied statistics using appropriate software (either MatLab or R, an open source (free) statistical software package and programming environment). Each lab session will be planned around a specific aspect of statistics and will require the preparation of a final report, with methodology, results and interpretations. Topics covered will include: Basic use of software, graphical data exploration and data summary, transformations. Simple linear regression, model specification, fitting, testing, model checking, interpretation. Multiple linear regression, model checking, variable selection, model comparison. Experimental design, response surface, paper helicopter design experiment. Quality control, action charts, components of variation
Module Resources
Applied Statistics for Engineers and Physical Scientists (3rd Ed) Ledolter, J & Hogg, R V. Pearson
Back