0

$Curve Model f(x) = ax^2 + bx + c$

$Data(x,y) = {(x_1,y_1),(x_2,y_2)...(x_n,y_n)}$

$Square Error = {E_1,E_2...E_n}$

SSE: sum of SE

$Initial values = {a=1 ,b=1, c=1}$

I'm new to statistics. In nonlinear regression, could I find best fitting coefficients by changing the initial values?

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
grape100
  • 1
  • 2
  • This is still a linear regression so we can do the normal ols equation or iteratively change the initial values to minimize the sum of squared error using gradient descent. – Tylerr Jul 14 '21 at 03:29
  • @Tylerr, Thanks for the tip. – grape100 Jul 14 '21 at 04:21

0 Answers0