42-AI / bootcamp_machine-learning

Bootcamp to learn basics in Machine Learning
Other
205 stars 41 forks source link

ML02 ex05 (mylinearregression): multiple issues (examples, syntax) #239

Open mli42 opened 2 years ago

mli42 commented 2 years ago

There are 4 issues on this exercice:


1) Init of MyLR

In the example, there is:

mylr = MyLR([[1.], [1.], [1.], [1.], [1]])

Thus giving a list of list, whereas in the previous module, we are given a np array:

lr1 = MyLR(np.array([[2], [0.7]]))

2) mylinearregression

In the previous module, the class was named my_linear_regression (with _), which, imo, is more readable, and it should have the same name for this module. It is currently not producing any bug, but it is a bit inconsistent.


3) mylr.theta

In the examples, it tries to access mylr.theta, but in the previous module the variable was named thetas with an S. A bit inconsistent too.


4) loss / loss elem

In order to satisfy the examples, I noticed the division by 2m is performed during loss() and not loss_elem(). Imo loss should be only the sum of loss_elem, as it was in the previous module.


Fixed on: