Classification in Python | logistic regression, LDA, QDA | Data Science With Marco

Notebook and dataset: github.com/marcopeix/datascie...
📚 Theory: 0:00 - 7:07
🐍 Code: 7:08 - 26:36
In this video, we cover the topic of classification in data science. We learn about logistic regression, linear discriminant analysis (LDA) and quadratic discriminant analysis (QDA), and use these algorithms to build a classifier for edible or poisonous mushrooms in Python.
Follow me on Medium: / marcopeixeiro

[@camerongridley9065]

Really enjoying your videos so far. Really clear and great code. Thanks for putting them together. A few suggestions for this one: 1) cover what to do if target class is imbalanced 2) pick a different target that doesn't produce a perfect ROC/AUC so we can actually see how different models perform. Keep up the great work!

[@datasciencewithmarco]

Thanks for the feedback!

[@priyadoesdatascience5141]

Thanks for the detail video, can you tell us how to test the assumptions for LDA and QDA using python

[@saswatprusty6691]

hey, can you show a case study where LDA is used over logistic regression or explain when to use one over the other?

[@rablaze]

how do you set y_pred_lda = np.where(np.where(y_prob_lda > .5, 1, 0) if you have more than two classifications?

[@datasciencewithmarco]

You do not necessarily need to set the class based on the probability. You can directly use the predict method of the LDA implementation in scikit-learn.

[@moak4052]

Hallo, can you make a video about resampling methods (bootstrap, cross validation

[@datasciencewithmarco]

Yes! The next video will be about resampling and regularixation!

[@pedroribeiro6271]

good video, still didnt use cross validation did you? You imported it but didnt use it right?

[@datasciencewithmarco]

Yes, you're right. I did not want to use cross-validation without explanation. So I made a video where I dive deeper into the subject: kzbin.info/www/bejne/mXi1aHSibJp2n8k Afterwards, I use cross-validation in all my subsequent tutorials! Cheers!

[@pedroribeiro6271]

@@datasciencewithmarco thanks for the answer :D