Coefficient of Determination definition
The coefficient of determination, R², measures how much observed response variation a fitted model accounts for relative to predicting the mean.
Coefficient of Determination in a picture
How it works
For an ordinary least-squares model with an intercept and a varying response, R² = 1 − SSE/SST. SSE is the sum of squared residuals. SST is the sum of squared differences from the response mean. In simple linear regression with an intercept, R² equals the square of the sample correlation r.
Worked examples
Find R² if SSE = 15 and SST = 60.
1 − 15/60 = 0.75.
For simple regression with intercept, r = −0.9. Find R².
(−0.9)² = 0.81.
Can a high R² establish causation?
No. It describes fit relative to the mean benchmark.
Common mistake
State which response variation the number summarizes. R² does not give the proportion of predictions that are exactly right.
Quick check
Can two fits have the same R² with slopes of opposite signs?
Show answer
Yes.
What does R² = 1 mean in this ordinary least-squares setting?
Show answer
The model has zero residual error on the fitted data.
What must be nonzero for SSE/SST?
Show answer
SST.
Sources
Flashcard
Coefficient of Determination
Reveal the definition
The coefficient of determination, R², measures how much observed response variation a fitted model accounts for relative to predicting the mean.
Effortless Math. “Coefficient of Determination.” Math Dictionary.
