Multivariate Analyses
Profile data allow also the following computations:

Principal Component Analysis

  • Metric: standardized (correlations), or identity (covariances).
  • Computation of missing data.
  • Means or individual scores taken in account. Additional individuals.
  • Detailed result edition.
  • Correlation circles.
  • Planes of the individuals and/or the means.
  • BiPlot.
  • Analyses may be done also by judge and by descriptor (consistency of the judges).
  • Horizontal analyses allow individualizing the attributes by judge. The variables are the Judge - Attribute pairs. Different weighting can be applied to the judges (MFA, STATIS ...).

Example of a BiPlot from Principal Component Analysis:

Example of a BiPlot from Principal Component Analysis

Optionally you may also project the individual judge scores, which was done in the example shown here (empty symbols). The points with the same given color represent the varied scores for a given product:

Example of a BiPlot plane from a PCA

Factorial Correspondence Analysis

  • Detailed result edition.
  • Graph of the lines and the columns.

Discriminant Analysis

  • Detailed result edition.
  • Correlation circles.
  • Graph of the individuals and the groups.

Ascending Cluster Analysis

  • Euclidean and Chi-Square distances.
  • Aggregation:
    - simple (minimum step),
    - complete (maximum step),
    - mean distance,
    - variance (Ward method),
    - centroid.
  • Detail of the classes.
  • Dendogram edition

Internal and external preference mapping

  • For the external mapping:
    Models:
    - vectorial,
    - circular,
    - elliptical,
    - quadratic.
  • Choice and validation of the model by consumer (AUTOFIT method).
  • Graphical representations:
    - consumer modeling,
    - consumer map,
    - consumer density.