0

I have 1200 row X 3000 columns . I have known that the parameters cannot be estimated when columns are larger in Linear Regression. Why does it work in Sklearn LinearRegression()?

desertnaut
  • 57,590
  • 26
  • 140
  • 166
Andrew Oh
  • 1
  • Where did you read this? Do you confuse with SVM? – Adrien Kaczmarek Jul 24 '20 at 16:31
  • This is actually a good question, and the answer appears to be undocumented (perhaps because it's abstracted away via SciPy's `lstsq`). I suspect it's using the pseudo-inverse when the matrix is rank deficient, which finds the minimum l2 norm coeffiicents – Nick Becker Jul 25 '20 at 02:15
  • https://stackoverflow.com/questions/23714519/how-does-sklearn-do-linear-regression-when-p-n – Nick Becker Jul 25 '20 at 02:18

0 Answers0