Shannon- Please review section 4

General Comments: 
This section is well organized, and clearly conveys the information about what was done, why, and what the results were. There are a few edits to make to the paragraphs, within section organization comments. But overall this looks like it's in very god shape. 

Specific Comments: 
Paragraph 1- The goal of of this paragraph is to introduce the high level idea behind empirical Bayes, and lead into the assumptions. From this, I think we could turn "Empirical Bayes" into the subject of the first sentence, contrasted with "The process behind empirical Bates". So we could re-write as: 

Empirical Bayes is an intuitive method. At a high level, EB assumes that future flu seasons look like past few seasons. [Sentence on Why we need assumptions] 

Paragraph 2: Point is that brooks made these assumptions, and we are looking to relax the first assumption. It seems like we could mention the purpose is to relax the first assumption before listing them.

Section 2.1, Overall: it seems like we should reverse the ordering around here. Maybe introcude Brooks' method, and these assumptions he uses. Then mention that assumption 1 is invalid, and we hope to relax it in this paper. That way, there will be a more complete story within the section, and it will lead better into 2.2.   

Sec 2.2, Paragraph 1: The purpose of this paragraph is to motivate why
the regional effects extension is necessary. From this, sentence one
could be changed to follow this storyline a bit better, something like
'Extending EB to include regional effects comes from our suspicion
that regions aren't independent from eachother. Then list the pieces
of evidence that follows.

Sec 2.2, Paragraph 2: Drop "Like EB, EB" just go with "Empirical Bayes
with Region Effects (EBRE)...". I think here, the fear is starting a
sentecne with an acronym.

Sec 2.2, P2: Reading this section independently, it's unclear what
"shifting and scaling" the wILI curve means. In terms of keeping the
writers cognitive load low, maybe mention: Recall the response
variable is the wILI curve, which measures the indcidence of the flu
over the entire season.

Sec 2.2, P3: I'm a bit unclear on "We ‘pin’ the values that are under
the flu threshold, meaning we do not adjust these values. The reason
being, that as the flu curve is not a linear function, scaling these
values that are already close to 0 distorts the analysis."  Perhaps we
could provide an example number here which would not be pinner.

End of Sec 2.2: I think you have reference figure 4 here a bit
earlier. Having this image in mind before going through the details
can help the reader to parse it.

Sec 2.2.1: This is a very clear subsection. The only thing I can point
out is that we could introduce lambda before the equation.

Sec 2.3, P3: I think "reasonable looking" to "reasonable". Overall,
this paragraph's goal is to display using the figure that the
predictions coming fromt he EB model are reasonable, and match with
our intuitions. I think this is a very solid paragraph.

Sec 2.3, P1: We could move these to the end of the section, maybe in a
single paragraph denoting the caveats.

Sec 2.4, P4: The first sentence, it seems like the term we are
introvuging is the MAE, and for this reason we can move this to the
stress position of the sentence. IE: Like the CDC, we measure the
quality of our predictions using the MAE.

Final paragraph; Maybe explain the reasoning of why we need peak
height, I don't think it follows as the obvious next step after
reading this. Maybe; Because predicting the entire curve lead to
errors, we want to simplify things and predict just the peaks, to see
if the better model works here.