Add Pop in NYCHA housing to Housing Security output

[AmandaDoyle]

Add the following data points to the Housing Security output.

• Total Population (Public Housing and RAD)
  • Count
  • Percent
  • By Race and ethnicity
• 2020 Census Total Pop values, which are in Demographic output. OSE cannot cross reference tables. 

Logic:

• Total Population (Public Housing and RAD) by race and ethnicity. Logic is to sum Public Housing White and RAD White, for example
• Denominator to calculate the percent is the 2020 Census Total Pop values, which are in Demographic output

Data source: Equitable Development Data Tool - Displacement Risk Index 2-8-2022 (1).xlsx This is used for other existing data points.

For reference,the lower box is where these data are being mapped

[td928]

@AmandaDoyle does the columns naming rad_pop and public_housing_pop make sense to you for those indicators?

[AmandaDoyle]

@td928 there is only one "one column" (i.e. nycha_tenants could be a good name) then nycha_tenants_anh for example. We need to sum the two columns Public Housing and RAD to get the total number of NYCHA tenants by race and ethnicity. Ping me if you'd like to hop on a call to talk this through.

[td928]

okay got it then! That's what you prompt actually says. I somehow thought we need to include the breakdown as well. Thank you! @AmandaDoyle

[SashaWeinstein]

I believe this is the first instance of us referencing a particular data point in two different categories? Curious to see how you do this Te. My first thought is to call the decennial census accessor and take the correct column from the dataframe it returns. Of course there are multiple good ways to do it

[td928]

yep this is my thought process too. The census accessor is pretty much what we need apart from the extra pct columns and easy to get rid off. Then hopefully the percent calculation for each race category is straightforward

[td928]

Question:The source data for this based off 2010 PUMA and since we are joining with 2020 census results there are a few discrepancies e.g. puma 03810, 03901 not available in spreadsheet provided. Is this expected and okay to leave as null? @AmandaDoyle

[AmandaDoyle]

@td928 thanks for flagging. I may be mistaken, but do we have a resource to map 2010 PUMAS to 2020 PUMAS? If so, we should use that. If not, then we can do a 1:1 join and leave fields where there is no match blank and flag it to OSE and leave it as a future improvement.

[td928]

Okay maybe I am not aware of it. Loop in @mbh329 and @SashaWeinstein for help, do we have a mapper from 2010 pumas to 2020 pumas?

[SashaWeinstein]

We haven't done anything with 2020 PUMAs. We looked at 2000 and 2010 PUMAs for our own ACS PUMS pipeline that we didn't end up using. Don't know anything about 2020 PUMAs, but could there be another reason they are left off source data?

[SashaWeinstein]

This site from dcp population maps old to new geographies. I remember talking with I think Joel about it but I don't remember what the conclusions were. It doesn't map from 2010 to 2020 PUMAs so maybe they are the same?

[td928]

Thanks for looking into it @SashaWeinstein. I also remember from past conversation they are mostly equivalent. So it is kind of interesting they are left off maybe for censoring purposes? But they are PUMA with population as low as 1 reported so I don't think there is a threshold that will not get reported. I am going with @AmandaDoyle suggestion now just treat them as non-report (null) and think of them as equivalent.

[mbh329]

@td928 did they leave off the pumas in the xlsx for public housing tenants because there just aren't aren't any nycha buildings?

[levysamu]

Hey all, I'm not sure why you are referencing 2020 PUMAs here. The NYCHA data is using 2010 PUMAs and the demographic tables are using 2020 census data mapped to 2010 PUMAs already. There should not be anything to calculate or convert. Let me know if we need to jump on a call quickly.

[td928]

Okay great @levysamu we had concerned about the population and NYCHA is not consistent on geographies. As long as they are all the same then it is perfect for our purpose and this should be ready to be review soon.

[td928]

okay this is now closed with #266 merged in and will have new output for SME to look at.