PCA
Real data often has many columns. PCA squeezes them into a few new columns and keeps most of the information. You can then draw the data, train faster and remove some noise.
New words
- PCA
- A method that turns many columns into a few new ones.
- Principal component
- A new direction that PCA finds. The first one has the most spread.
- Variance
- A number that shows how much the values spread out.
- Projection
- Dropping each point straight onto a line, like a shadow.
- Covariance matrix
- A small table that shows how every pair of columns spreads together.
- Explained variance
- The share of the total spread that a component keeps.
What you will learn
- Can many columns become one?
- Why would you want fewer columns?
- Which direction has the largest spread?
- How does the covariance matrix help?
- What are principal components?
- How do you project the points onto the line?
- Try it: rotate the line and watch the shadows
- What about 5 subjects?
- What is lost, and when does PCA mislead?
- Common mistakes
- Explain it to a friend
- Key takeaways
10 quiz questions at the end.
The full lesson, with interactive demos and the quiz, opens after you sign in with Google. It is free.