lesson03 script notebook issues

[elcorto]

I'm going through the material and will for now just list things I find in https://github.com/deeplearning540/deeplearning540.github.io/blob/main/source/lesson03/script.ipynb

• [2] why not use dropna as in lesson02?
• [5] I think is is confusing to call the KNN classifier kmeans, which was just introduced as a unsupervised method (clustering) in the previous lesson
• [49] and [36] (below that): because 1 = true and 0 = false, the binary confusion matrix

  [[tn, fp],
   [fn, tp]]
  

  output by sklearn doesn't match the figure from Wikipedia, which is confusing
• [49] and [36] (below that): the confusion matrix convention in sklearn is axis 0 (rows) = true, axis 1 (cols) = pred, the Wiki image has axes swapped
• [50]: print(int(True)) ?
• [54]: F1 score formula TeX error (ctimes)

[psteinb]

on [2]: indeed, but I wanted to show multiple ways to reach dropping the NA

[psteinb]

on [5]: fair enough, I need to watch out for this

[psteinb]

[36] and [49]: I know, this is something I comment on in the video.

[psteinb]

[50] was needed in older versions of matplotlib 🤷

[psteinb]

[54]: please submit a PR ... not sure I have the time to fix this

[elcorto]

on [2]: indeed, but I wanted to show multiple ways to reach dropping the NA

ah, ok. Still I think dropna is less confusing, more common and in sync w/ the previous lesson

[elcorto]

[54]: please submit a PR ... not sure I have the time to fix this

Yes same here :) I need to go over all material at least once before the next course, that's why only the quick list for now.