This is an example, in excel, where I try to find parameters of a linear regression. (Correction: The error is not well computed. Need to add the squares of G15-G20.)
A question if I may, referring to the formula, at 8:26 aren't you supposed to minus theta1 with learning rate / size of training set * sum of the whole G15 to G20 (instead of just G15)? Thanks.
Calculation of gradient descent requires derivate of (h(theta) - y)^2. May I know why didn't you calculate the derivative or is it just not used in practice?
A question if I may, referring to the formula, at 8:26 aren't you supposed to minus theta1 with learning rate / size of training set * sum of the whole G15 to G20 (instead of just G15)? Thanks.
Calculation of gradient descent requires derivate of (h(theta) - y)^2. May I know why didn't you calculate the derivative or is it just not used in practice?
It is baked in the algorithm.
what is the formula to reach error sq. ??
I think he has it wrong. Average of the sum of squares should be 68639.667 in the first iteration.
aren't we suppose calculate THETA1 and THETA2
initially using linear regression coefficient formula.