Shape of cost function for multiple variable linear regression?


In C1 W1, professor mentions that for “Single Variable Linear Regression” where
f = wx + b, the cost function if we choose squared error loss, is in the shape of a bowl (convex function).

In C2 W2, we learn about “Multiple Variable Linear Regression” where
f = w1x1 + w2x2 + … + wnxn + b
What is the shape of the cost function in this case if we use squared error loss? Will it still remain like convex function.

It is still convex, but it cannot be plotted visually because it is multi-dimensional.

1 Like