Principal Component Analysis (PCA) Tool screenshot

Principal Component Analysis (PCA) Tool

Run a two-variable PCA on pasted rows: means, covariance, both eigenvalues, variance share and a scatter with the component axes.

Privacy First
0 Runs
Updated 2026-11-06
No upload
Runs in this tab

Two columns, two components

Paste one x, y pair per line. The page computes the column means, the sample covariance matrix with the n − 1 denominator, and both eigenvalues in closed form for a symmetric 2×2 matrix: λ = T/2 ± √(T²/4 − D), where T is the trace and D the determinant. It prints the eigenvalues, the share of variance the first component carries, the angle of that component and a scatter plot with both axes drawn through the mean.

This is PCA for two variables, so there is nothing to iterate and nothing to converge. It does not standardise the columns — the covariance is on the raw units, and a column measured in thousands will dominate one measured in fractions — and it does not reduce a larger table, whiten the data or score new rows. Fewer than three rows is refused, because two points describe a line rather than a spread.

The means and the decomposition are computed in this tab from the pasted text. Nothing is uploaded, the dataset is not saved, and the same rows print the same components after a reload.

Run a two-dimensional principal component analysis on pasted rows: the covariance matrix, both eigenvalues, the PC1 angle and each component's share of the total variance.

The covariance uses the sample denominator n − 1, so the shares describe the spread of the data you pasted — computed in this tab and never uploaded.

How to use it

Paste two numeric columns, one row per line, with commas, tabs or spaces between them. The page needs at least three complete rows, builds the covariance matrix and decomposes it, then draws the scatter with the mean and the first axis.

The arithmetic

The covariance matrix is symmetric 2 × 2, so its eigenvectors are the principal axes and the eigenvalues give each component's share of the total variance. The answer prints PC1's angle from the x-axis, the mean point and both shares.

Where it stops

Two variables, at least three rows, no missing values and no rotation beyond the two components. A set where every point is identical has no variance to decompose, and a row that is not two numbers is refused with its line quoted.

What to Expect

Best for

  • seeing which direction a two-variable cloud spreads in
  • checking a covariance computation by hand

What this tool keeps

  • The covariance, the eigen-decomposition and the drawing all happen in this tab; the rows are not stored.

Frequently asked questions

Which denominator does the covariance use?

n − 1, the sample denominator. The page says so on the controls, because with n it would report a different variance share for the same points.

Why does it need three rows?

Because two points define a line, not a spread to decompose. Three is the minimum where a covariance matrix has something to say.

What does the variance share mean?

The fraction of the total variance carried by each component. With two variables the two shares add to 100%, and the angle says which direction the larger one points.

Developer program

Build custom tools for your store

Join the SDK program to create useful ecommerce utilities for your store.