43859 (3) [Avatar] Offline
This is really small. In chapter 4, in the code samples, there are some imports missing: pd (pandas), TruncatedSVD (from sklearn.decomposition ?), train_test_split (from sklearn.model_selection ?), LDA (not sure about this one, is it sklearn.lda ?).

Another small thing, in section 4.5.5 where you compare the topic vectors obtained from TruncatedSVD and PCA, you proceed to perform the cosine similarity for svd vectors. Is the cosine similarity for pca vectors missing, or did you intend it as an exercise for the reader?