Enter PCA Inputs
Paste eigenvalues or per-component variances from your PCA output. The calculator sorts them in descending order automatically.
Example Data Table
This example uses eight principal components with a total variance of 10.0.
| Component | Eigenvalue | Explained Variance | Cumulative Variance |
|---|---|---|---|
| PC1 | 3.80 | 38.00% | 38.00% |
| PC2 | 2.40 | 24.00% | 62.00% |
| PC3 | 1.50 | 15.00% | 77.00% |
| PC4 | 0.90 | 9.00% | 86.00% |
| PC5 | 0.60 | 6.00% | 92.00% |
| PC6 | 0.40 | 4.00% | 96.00% |
| PC7 | 0.25 | 2.50% | 98.50% |
| PC8 | 0.15 | 1.50% | 100.00% |
Formula Used
EVRi = λi / Σλ × 100
CVk = (Σ λi from i=1 to k) / Σλ × 100
Reduction % = (1 - k / p) × 100
Compression factor = p / k
Residual % = 100 - CVk
deff = (Σλ)2 / Σ(λ2)
κ = λ1 / λk
SRR = (selected variance) / (discarded variance)
Here, λ represents each eigenvalue, k is the number of retained components, and p is the original number of features.
How to Use This Calculator
- Run PCA in your analytics software and copy the eigenvalues or component variances.
- Paste them into the input box using commas, spaces, or line breaks.
- Enter the original feature count and optional sample count.
- Set a target cumulative variance, such as 90%, 95%, or 99%.
- Leave manual components blank for automatic selection, or enter your own retained count.
- Choose whether the source variables were standardized before PCA.
- Submit the form and review retention, reduction, scree shape, and component-level detail.
- Use the export buttons to save the summary and component table as CSV or PDF.
FAQs
1) What does this calculator evaluate?
It evaluates how well a chosen number of principal components preserves variance, reduces dimensionality, and balances compression against information loss. It also highlights scree behavior and stability-related metrics.
2) What should I paste into the eigenvalue box?
Paste the eigenvalues or component variances produced by your PCA routine. You can separate numbers with commas, spaces, semicolons, or line breaks.
3) Does the order of eigenvalues matter?
PCA interpretation assumes descending order, but the calculator sorts the values automatically. That helps prevent mistakes when data is pasted from mixed outputs.
4) When is the Kaiser rule useful?
The Kaiser rule, which keeps components with eigenvalues at least one, is most meaningful when PCA is based on standardized variables and a correlation-style scale.
5) What target variance should I choose?
Common targets are 90%, 95%, and 99%. Lower targets compress more aggressively, while higher targets preserve more information but keep more components.
6) What does the scree elbow mean?
The elbow marks where additional components begin contributing much smaller gains. It is a visual guide, not a strict rule, so combine it with variance thresholds and domain judgment.
7) Why can high compression still be risky?
A very small retained set may compress strongly yet discard important structure. Always compare reduction percentage with residual variance, model goals, and downstream performance.
8) Can I compare two PCA runs with this tool?
Yes. Run the calculator separately for each PCA output and compare retained variance, effective dimension, elbow location, and compression factor to judge which solution is more practical.