Service Analytics

6.Principal Component Analysis

Popular Stopping Rules*

  • Kaiser-GuttmanCriterion:
    • lampda >1
  • Scree Plot:
    • plot the eigenvalues of the principal components against their rank: --> "Ellenbogenkriterium"
      • the break between the steep slope and a leveling off indicates the number of meaningful components
  • Proportion of total variance:
    • keep principal components until a certain percentage of total variance is reached, e.g. 95%

Diskussion