Folder:
113 Math
File:
113.034 Statistics - Coefficient of determination and Squared error

Coefficient of determination and Squared error

r^2 = coefficient of determination
- Always between zero and one: 0 >= r^2 <= 1
- r^2 is literally just r squared, it's that easy
- $$r^2 = 1 - \frac{SE_{regression}}{SE_\bar{y}}$$
- Reads: r squared = 1 minus the squared error of the regression line divided by the squared error of the average of all y values
- If squared error is small, it means that the residuals of all points along the regression line are small and they fit the line pretty well
- If squared error is big, it means that the residuals of the points along the regression line are big and they don't fit the line very well
- So, r^2 close to 1 is a good fit and r^2 close to zero is a bad fit


Tags:
  • statistics