You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Your proposed structure is a lot closer to how we organised our Y4 course last year. After this we decided on the new structure: why? We think functions should go early because otherwise you are using print() etc without having them properly explained. Ditto for modules using e.g. math.sqrt(). numpy arrays can be introduced as “special” collections. And you can do realistic exercises without using any logic or flow control and still understand everything, so these go last.
That said, this is not fixed in stone, but is how we are planning to structure the Y4 course next year.
Building on Ben's comment, I spoke with João Malhado from Imperial, who mentioned that their Python course (focused on data analysis) doesn't ever mention loops as the NumPy vectorisation renders them unnecessary. So I think that there is some method to our madness of having a non-traditional order.
Obviously this may all change when we actually have content, so I think we should keep this issue open for now.
Activity
bjmorgan commentedon Apr 2, 2020
Your proposed structure is a lot closer to how we organised our Y4 course last year. After this we decided on the new structure: why? We think functions should go early because otherwise you are using
print()etc without having them properly explained. Ditto for modules using e.g.math.sqrt(). numpy arrays can be introduced as “special” collections. And you can do realistic exercises without using any logic or flow control and still understand everything, so these go last.That said, this is not fixed in stone, but is how we are planning to structure the Y4 course next year.
arm61 commentedon Apr 3, 2020
Building on Ben's comment, I spoke with João Malhado from Imperial, who mentioned that their Python course (focused on data analysis) doesn't ever mention loops as the NumPy vectorisation renders them unnecessary. So I think that there is some method to our madness of having a non-traditional order.
Obviously this may all change when we actually have content, so I think we should keep this issue open for now.
bjmorgan commentedon Apr 3, 2020
No loops or logic in our Y1 labs either, and they are kind of shoe-horned in to the Y2 lab. You could build the same model without using either.
j-m-dean commentedon Apr 4, 2020
Fair enough, I guess I am biased by my own python education wherein numpy vectorisation wasn't even taught in depth until the second year course.