C1_W2_Lab03, the alpha that results in faster convergence

[RushIbrahim]

The screenshots are from the optional lab about multivariate feature scaling from Week 2.

The text below the results from the second alpha mentions that it doesn’t converge as quickly as the previous example, but doesn’t doesn’t alpha two (alpha = 1e-7) actually decrease the cost much quicker (when comparing their respective costs after each 10, 50 or 100 iterations)?

alpha one:

alpha two:

[rmwkwok]

Hello @RushIbrahim, good observation! However, if you choose 200 iterations, you can see that the first example will overtake.