Create an overview doc of Ex. 21 formats/layouts

[katie-lamb]

As a first step in the extraction process we need to scope the different formats that the Ex. 21 documents come in so we can have a better understanding of what layouts to try to extract from first. Create a document with the following:

Tasks

Preview Give feedback

1. Target variables and which ones we can access through each layout
2. Screenshots of each common layout
3. Rough distribution of documents to each layout
4. Experiment with off the shelf extraction tools
5. List modeling method pros and cons and create issues for model experiments to try
6. Create issue for extracting company name and address information from 10K's (main filing not Ex. 21)
7. Figure out what the difference is between different quarters of filings. What's in the Q1 filing that is not contained in the other quarters?

Once we've conducted this overview, we'll have a better idea of how to create a representative sample of documents to begin extracting from, or which layout to start with.

[katie-lamb]

Began creating a google doc to track this, so far main issue is that sometimes Ex. 21 tables are within an HTML table while other times they're part of a HTML body. Even when using pandas read_html to read in from an HTML table there needs to be some serious reformatting to get the schema right. This might require categorizing into the different formats/schemas.