Overview
Multivariate methods analyse multiple variables simultaneously and uncover latent structures, natural groups or classification rules in datasets. The four calculators available here cover the most important use cases – from dimensionality reduction to supervised classification.
Interaktive Rechner
Principal Component Analysis
Eigenvalue decomposition of the correlation matrix, Scree plot, Biplot, optional Varimax rotation and Kaiser criterion.
Factor Analysis
Principal axis factoring (PAF) and PCA extraction, Varimax rotation, KMO measure, Bartlett test and communalities.
Cluster Analysis
Ward, Complete and Average Linkage with dendrogram and k-Means with Elbow method and Silhouette coefficient.
Discriminant Analysis
Lineare Discriminant Analysis (LDA), Wilks' Lambda, kanonische Korrelation, Konfusionsmatrix und LOO-Kreuzvalidierung.
Method comparison
| Criterion | 📐 PCA | 🔬 Factor Analysis | 🔵 Cluster Analysis | 🎯 Discriminant Analysis |
|---|---|---|---|---|
| Objective | Dimensionality reduction | Latent factors | Group formation | Group classification |
| Groups known? | – not required | – not required | – not required | ✓ required |
| Objectivevariable | keine | keine | keine | Group membership |
| Rotation | Varimax (optional) | Varimax / none | – | – |
| Goodness criterion | Explained variance | KMO, Bartlett | Silhouette, WCSS | Wilks' Λ, Hit rate |
| Output | Scores, Loadings | Loadings, communalities | Cluster assignment | Classification, DF scores |
| Typical application | Index construction, preprocessing | Scale construction, psychometrics | Segmentation, taxonomies | Credit scoring, diagnostics |