PCA
PCA (noun) means Principal component analysis: a statistical technique that reduces a large set of correlated variables to a smaller set of uncorrelated components that still capture most of the variation in the data. Example: “We ran a PCA to cut sixty survey items down to four underlying factors.”
How to Use PCA
Learner’s notesIn plain EnglishMost often it means a maths method for squeezing lots of data down into a few key numbers.
Always define the expansion on first use in writing, since the three main meanings sit in very different fields.
In data science contexts PCA nearly always means principal component analysis; in a hospital it nearly always means the pain pump.
Trace the full origin ↓Fill the Gap
Can you complete this real example?
We ran a _____ to cut sixty survey items down to four underlying factors.
Etymology
An initialism, expanded differently in different fields.
Frequently Asked Questions
What is the difference between PCA and factor analysis?
PCA is descriptive: it repackages the observed variance into components with no assumption about what causes it. Factor analysis is a model: it assumes hidden latent factors generate the observed variables and estimates those factors. They often give similar-looking output but answer different questions.
What does a PCA pump do?
It delivers a fixed small dose of painkiller when the patient presses a handset, then locks out further doses for a set period so an overdose is not possible. Staff programme the dose size, lockout interval and total hourly limit.
How many principal components should you keep?
There is no single rule. Common approaches are keeping components that together explain a target share of variance, keeping those with eigenvalues above 1, or reading the elbow on a scree plot. The choice should also be guided by whether the retained components are interpretable.