Script to amend monthly rollup microdata

[nmdefries]

Description

Add state field based on county FIPS. Change name of wave field to
version.

Changelog

• Add script microdata_add_state_col__rename_wave.R

[capnrefsmmat]

I think this will be fine, just slightly more complicated than it has to be, since the FIPS conversion can be easier.

Not sure if it's worth changing the conversion if it already works. Could you maybe add an assertion after line 42 that checks that all rows with FIPS codes get a non-NA state? That should definitely happen, but if there's something weird about our mapping files (like with territories), we could get issues, and we don't want to have missingness because of that.

[nmdefries]

The logic for these two functions is borrowed from delphiFacebook/R/responses.R, but I'm wondering if this line is supposed to be handling A3. We drop zip5 anyway, so nulling it out here seems unnecessary. @capnrefsmmat

[capnrefsmmat]

filter_complete_responses does the other half of the work:

covidcast-indicators/facebook/delphiFacebook/R/responses.R

Lines 772 to 777 in 3590e64

	# what zip5 values have a large enough population (>100) to include in micro
	# output. Those with too small of a population are blanked to NA
	zip_metadata <- produce_zip_metadata(params$static_dir)[, c("zip5", "keep_in_agg")]
	zipitude <- left_join(data_full, zip_metadata, by = "zip5")
	change_zip <- !is.na(zipitude$keep_in_agg) & !zipitude$keep_in_agg
	data_full$A3[change_zip] <- NA

We join with the ZIP metadata based on zip5, then blank out A3 based on the results (not zip5). Don't ask me why we need to use two different columns; undoubtedly the code did that in early 2020 and the logic just got ported over to this version

[nmdefries]

@krivard This is ready to merge.