for info: if you delete the column island then you should delete the rows containing value 1 as well or you will have the other encoded columns equals zero in all.
i've been lost in feature engineering chapter on the book that i am currently ready for machine learning right now, and straight ahead i found your video, and with the whole 47 minutes i have learned 2-3 things from you and i understand the whole process lot more better now, this all thanks to you Greg! keep up making these types of videos bcs WE NEED YOU!!
at 4:30 after looking it up, "frac=1" sets the fraction of the dataset to shuffle to be 100% of the total dataset, instead of say 25%. didn't really explain that properly
Wouldn't this be called Transformation techniques of preprocessing instead? I thought Dimensionality Reduction would be separate from Feature Engineering, with Feature Scaling making up the 3rd subtopic. So something like: Dimensionality Reduction (removing features) PCA Clustering Feature Engineering (creating/transforming features) One-Hot Binning Feature Scaling (normalisation/standardisation of features) Your scaling
The techniques you showed were very important but don't know why you were ruining it by training on illogical models. I understand that it was just for test purpose but you could do better by training some logical models.