A training course developed jointly by Hal Macfie & Anne Hasted
This course forms a hands-on introduction to the multivariate statistical methods needed by a sensory scientist. Mainly based around the use of SENPAQ and XLSTAT software, attendees will gain practical experience using real-life examples and will be able to save their work to take away with a free copies of SENPAQ and XLSTAT so they will be competent to start using their expertise from day 1 after the course.
ABOUT THE LECTURER
Anne Hasted began her career as an academic statistician at Reading University but now runs Qi Statistics Ltd which offers statistical training, data analysis and consultancy support to research and industry and has developed both SENPAQ and Design Express software packages. Anne has wide training and consultancy experience in both sensory and consumer research and has an international reputation for “user friendly” training.
| 09.00 - 12.30 |
Statistics refresher!
Quick refresher of basic statistical concepts
Analysis of sensory descriptive data |
|
Revision of QDA approach |
|
Comparisons of two samples -Student t-tests, Analysing QDA data using analysis of variance |
|
Testing for sample differences, least significant difference, multiple comparison tests, testing for interactions. Graphical displays |
|
Exercises using SENPAQ and XLSTAT |
|
|
| 12.30 - 13.30 |
Lunch |
|
|
| 13.30 – 15.00 |
Further Analysis of Variance
Random and mixed models, Analysing more complex factorial sample structures – Testing model assumptions |
|
|
|
Tea |
| >15.00 - 16.30> |
Panel Monitoring |
|
Discrimination, Reproducibility, Measuring consistency -partitioning the interaction effects, assessor vs panel mean correlations . Understanding SENPAQ output |
|
Setting up a Traffic light system for performance monitoring |
|
Exercises in SENPAQ |
| 16.30 - 17.30 |
Non Parametric Testing |
|
How to analyse data in the form of ranks-applications of non parametric methods when ANOVA assumptions fail. Exercises using SENPAQ and XLSTAT. |
|
|
|
|
| Evening |
Dinner hosted by Qi Statistics |
|
| Day 2 |
|
|
| 09.00 - 10.30 |
Principal Components Analysis of QDA data |
|
Covariance, Correlation, Interpreting Correlations using cosines |
|
Geometric interpretation of Principal Components Analysis |
|
PCA on Sensory data - covariance or correlation? |
|
Methods to determine dimensionality |
|
Effect of Outliers, Varimax rotation
Use of supplementary variables and observations |
|
How is PCA different from Factor Analysis? |
|
Exercises in SENPAQ/XLSTAT |
|
|
|
Coffee Break |
|
|
| 11.00 - 12.30 |
Cluster Analysis, clustering of products and variables to select key sensory attributes |
|
Similarity/dissimilarity measures |
|
To standardise or not? |
|
Agglomerative Hierarchical Classification (AHC) |
|
Aggregation criteria |
|
Clustering Products and variables |
|
Exercises in XLSTAT |
|
|
| 12.30 – 13.30 |
Lunch |
|
|
| 13.30 - 14.30 |
Multivariate methods to assess Panel Performance and Flash profiling |
|
Generalised Procrustes Analysis (GPA) |
|
Application to assess panel performance data, Flash profiling definition and analysis
Exercises using XLSTAT MX |
|
|
| 14.30 - 15.30 |
Multiple Factor Analysis |
|
Visualising data from different tables, RV coefficient, applications in linking sensory, analytic and consumer data
Exercises using XLSTAT MX |
|
|
|
Tea Break |
|
|
| 16.00 - 17.30 |
Canonical Variates Analysis, Discriminant analysis |
|
Discriminant Analysis (DA) |
|
Canonical Variates Analysis (CVA) |
|
Graphical Displays, Confusion matrices |
|
Exercises in SENPAQ |
|
| Day 3 |
|
|
| 09.00 - 10.30 |
Relating to instrumental and consumer data 1 |
|
Linear regression, Multiple regression |
|
Modelling Curvature, Stepwise regression |
|
Best modelling, Overfitting |
|
Principal components regression |
|
Exercises in XLSTAT |
|
|
|
Coffee Break |
|
|
| 11.00 - 12.30 |
Relating to instrumental and consumer data 2 |
|
Partial Least Squares regression compared to PCA regression |
|
Geometric interpretation |
|
Assessing dimensionality using Cross validation |
|
Relating sensory to instrumental data |
|
Relating Liking to sensory data |
|
Exercises in XLSTAT PLS |
|
|
| 12.30 – 13.30 |
Lunch |
|
|
| 13.30 - 14.30 |
Sorting and Napping |
|
What is Mutidimensional Scaling
Analysing distance and similarity matrices
Sorting trials.
Napping – methods and analysis
Exercise using XLSTAT |
| 14.30 - 15.30 |
Bayesian network Analysis
Introduction and applications to sensory data analysis |
|
|
|
|
| 16.00 |
Close |
Located in the central business district at 108 Robinson Road, Cliftons Singapore is just a few minutes walk from the Tanjong Pagar MRT station, all-day parking and historic landmarks like Raffles Place and Chinatown.
Fees reduced by 25% for members of academia - space limited.
Includes lunches and a course dinner, course materials, attendees receive a free one year licence to XLSTATpro and the sensory module XLSTAT MX and a permanent licence for SENPAQ (the specialised sensory analysis program developed for the course).
The combined value of these software programs is $US730 (it is possible for course attendees to convert to a permanent copy of XLSTAT and MX for a further payment of $US180-recommended as the full price is $1090).
Equally if you do not want the XLSTAT licence we can reduce the course fees by $US200.
Attendees may bring along their own PC laptops for the exercises and we will install a free copy of XLSTAT2007 and SENPAQ on your computer. (We do not guarantee to load XLSTAT if you have an unusual software configuration.)
