Additional to the comments below, you can use dropdown to hide:

• The Yeo-Johnson and Box-Cox transforms (as two separete dropdowns) in 6.3.2.2. Mapping to a Gaussian distribution. Keep the paragraph saying "Below are examples of Box-Cox and Yeo-Johnson applied to various probability distributions [...]" and the example image out of the dropdown.
• The Sparse input paragraph in 6.3.3. Normalization
• Create a dropdown for the paragraphs on Support of categorical features with missing values just above 6.3.4.1. Infrequent categories
• Create 3 dropdowns inside the 6.3.4.2. Target Encoder section: For the binary classification target, the multiclass target and continuous targets.

We are almost there! Here's another batch of comments.

Also, for creating the binary classification target dropdown in section 6.3.4.2. Target Encoder you can split the paragraph in current line 907, i.e. starting from "For the binary classification target, the target encoding is given by: [...]"

Just a couple of suggestions, but otherwise it LGTM :)