I just finished the ML class from Georgia Tech as part of the OMSCS program. I used SciKit for most of the assignments as they involved NNs, DT, KNN, K-means, EM. This might be a naive question as I'm not a python guy but is there a reason this book is python based but doesn't cover scikit-learn? For example, what need did you see to write code for k-means[1] than to use an implementation already available [2]

[1] https://github.com/joelgrus/data-science-from-scratch/blob/m... [2] http://scikit-learn.org/stable/modules/generated/sklearn.clu...