In this course on machine Learning by Stanford, Andrew Ng describes a function for a line as h(x) = Theta subscript 0 + Theta subscript 1 x instead of the more common slope-intercept form y = mx + c.

Thanks to a previous question, I now understand how both forms represent a line (via the y intercept and slope) but am unclear why Andrew Ng decided to use the Theta form of the equation rather than the more common slope-intercept form.

Can anyone help me understand this please?

  • 2
    $\begingroup$ $h(x)=\theta_0+\theta_1x$ is a common slope-intercept form. He just decided to name the parameters $c$ and $m$ differently. $\endgroup$ Sep 2, 2018 at 9:20
  • $\begingroup$ I've never done Ng's ML course, but I'm guessing it is because that notation is commonly used in the context of gradient descent and cost functions. $\endgroup$
    – Git Gud
    Sep 2, 2018 at 9:31
  • 1
    $\begingroup$ notation like $\theta_0+\theta_1 x$ generalises much better than $c+mx$ to the situation where there are multiple variables, which is going to be of interest in ML. But really it's a purely notational choice with no meaning. $\endgroup$ Sep 2, 2018 at 9:31
  • $\begingroup$ Someone downvoted this question. Any ideas why? It was a lefitimate question and the answers have helped me enormously. So I am curious if I did something contrary to protocol here? $\endgroup$
    – Axle Max
    Sep 3, 2018 at 17:46


You must log in to answer this question.

Browse other questions tagged .