Hi @lhighzer ,
Which part of the question you find it wired?
specifically the 2 part âwhen the lineâŚâ it s just different syntax holding the same meaning of the 1 part âexplain whyâŚâ leaving no room for the answer.
Perhaps you meant âweirdâ instead of âwiredâ?
Hi @lhighzer ,
I have no access to your quiz, so i can only based on what you posted here.
Depending on the cost function used, I assumed it is the common one, MSE, mean squared error:
J(w) = \frac{1}{m}\sum_{i=1}^{m} (f_w(x^i) - y^i)^2
f_w(x^{i}) is the model prediction of given input data i
and y^i is the true value of training data i
f_w(x^i) - y^i would give us the error or vertical distance of the line from the real data point
The J(w) formula collects all individual errors, squares them and average them across the entire dataset. Giving us a single number as penalty to work on. If that number is at its lowest value, that means your modelâs predictions match the real data as closely as mathematically possible.
As the question begins with âopen-ended:â, it might be asking for a written answer.
yeh âweirdâ* my bad
yeh obviously,but the structure of the question that I m discussing as not as it should be as I explained in the first reply.
Could you be clear what you are referring to? What do you see this question should be? give an example?
I agree entirely. The question is written in a confusing manner.