Principal Component Analysis (PCA) is a fundamental multivariate data analysis technique used to reduce the dimensionality of data while preserving as much variance as possible. This tool walks you through the full PCA workflow — from data entry and preprocessing through to interactive visualisation of scores, loadings, and biplots.

How to Use the PCA Explorer

The wizard guides you through four steps:

1. Data Input

Select one of the built-in example datasets or enter your own data using the editable table. The Simple dataset (Height, Weight, Age) is ideal for learning the basics. The Iris Subset and Chemistry examples include group labels for coloured score plots. Toggle Auto-scale (unit variance) to mean-centre and scale each variable — this is recommended when variables are on different units or scales.

2. Preprocessing

Review summary statistics (mean, standard deviation, min, max) for each variable. The Correlation Matrix is displayed as a colour-coded heatmap — dark blue indicates strong positive correlation, dark red indicates strong negative correlation. The covariance matrix is also shown below.

3. PCA Results

The Scree Plot shows eigenvalues (bars) and cumulative explained variance (green dashed line). Use this to determine how many principal components to retain — a common rule of thumb is to keep components with eigenvalues > 1 or until 70–80% of variance is explained. The Loadings table shows how each variable contributes to each principal component, and the Scores table shows the transformed coordinates of each observation.

4. Visualisation

The Score Plot displays observations projected onto the principal components you select. Points can be coloured by group if the dataset includes group labels. Enable the biplot overlay to see how the original variables contribute to the PC directions. The Loading Plot shows the contribution of each variable to the selected PCs as vectors. Export the score plot as SVG, download the full results as CSV, or copy a summary to your clipboard.