Cluster Analysis in Practice – Applications to Consumer Product Research Trials illustrated using XLSTAT

This live session gives you plenty of opportunities to ask Anne the questions you have had for so long! As well as the live lectures and discussions with Anne you will receive workshops in XLSTAT with annotated solutions – great for future reference!

This half day workshop starts with the two most widely used methods of clustering consumers based on their hedonic liking scores. The importance of evaluating the derived clusters is shown in a variety of ways. We then outline some of the recent developments to support clustering of consumers using other data structures typically collected in product research trials and consider how they may be used to enhance liking clusters.

• Agglomerative Hierarchical Classification (AHC) and non hierarchical K-MEANS
  • Applicability of each method
  • To rescale or not to rescale?
  • What makes a good solution?
  • Comparison with other methods available: Fuzzy Clustering and Latent Class Clustering
• Critical Evaluation of the solution
  • Identifying atypical consumers within a cluster – cluster trimming using correlation or Silhouette scores
  • Presentation order effects – can they bias a solution? How to deal with them
  • Cluster visualisation using PCA
  • Do clusters show different sensory drivers of liking ( from sensory panel data or consumer sensory evaluation)?
  • Can clusters be explained by differences in consumer demographics? Investigation using Chi-Squared tests/ Correspondence Analysis/ Discriminant Analysis
• Clustering using other responses typically collected
  • Using XLSTAT routine CLUSCATA – interpretation of output and cluster evaluation
  • Other methods proposed
  • Linking clusters based on CATA data with hedonic response clusters – how similar are they?
  • Clustering using other multivariate responses – JAR scales and data where each respondent has scored different variables: eg Free Choice Profiling and Napping. This can be done automatically using the XLSTAT routine CLUSTATIS or using other data reduction techniques such as MFA (Multiple Factor Analysis) or GPA (Generalised Procrustes Analysis

Pre-requisite for course:

The workshop assumes a basic knowledge of clustering methods and data mapping using principal components. You can up your skills on these techniques using our On Demand Modules C3 and C4 before the workshop

Software used: XLSTAT (see below)

Booking closes Thur 18th June 2026