Add FAQ

[gsheni]

• Added a notebook that goes through solutions for FAQs

[codecov]

Codecov Report

Merging #620 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #620   +/-   ##
=======================================
  Coverage   97.42%   97.42%           
=======================================
  Files         118      118           
  Lines        9539     9539           
=======================================
  Hits         9293     9293           
  Misses        246      246

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0c8f506...6c7549e. Read the comment docs.

[ctduffy]

2 comments:

1. Near code snippet 32/in section "How do I prevent label leakage with DFS":
   What about if the user uses train test split to split the training and testing data? This might be a better example, because there are a few things that must be done to ensure accuracy with this method (ie matching the labels after making the features). I also don't really understand fully how the training and testing data is generated/divided in this example.
   Also, how are categorical variables encoded when splitting training and testing data? What happens if the categorical variable has a value in the testing data but doesn't have it in the training data (this might be an uninformed question)

2. Below code snippet 64/in section "Why am I getting...?"
   Would it be possible to further clarify the difference between last time index and time index/clarify what it does with the excluded rows? Are these rows used in testing, or how are they factored in (if at all)? Again, not sure how well informed this question is.

These two questions might not be valid questions, in which case they can be ignored.

[kmax12]

LGTM