NYCPlanning/labs-factfinder

Decennial: homeowner vacancy rate and rental vacancy rate

Closed this issue · 9 comments

Data are not accurately appearing in the application. Possible fix is that the calculations are missing to piece to be multiplied by 100.

image
https://popfactfinder.planning.nyc.gov/profile/802/census

Does this issue prevent the new ACS data release? It seems like it's happening on production.

Okay, when I multiply the rate by 100, I get the follow data for the variable:

id: "SG1Pd25yVmNSdA==",
sum: 2.8832683294351167,
variable: "hmownrvcrt",
variablename: "HmOwnrVcRt",
base: "ratio",
category: "housing_occupancy",
profile: "decennial",
percent: null,
percent_m: null,
previous_sum: 1.6816467291734023,
previous_percent: null,
comparison_sum: 2.88326832943512,
comparison_percent: null,
change_sum: 1.2016216002617144,
change_percent: 0.7145505529882369,
change_percentage_point: null,
change_percentage_point_m: null,
difference_sum: -3.1086244689504383e-15,
significant: true,
difference_percent: 0

However, I'm concerned that this will affect downstream calculations that depend on it... there is one variable that reference this as its base, and this is what it computes to:

id: "T09jSFUx",
sum: 962892,
variable: "oochu1",
variablename: "OOcHU1",
base: "HmOwnrVcRt",
category: "housing_occupancy",
profile: "decennial",
percent: 0,
previous_sum: 912296,
previous_percent: 0,
comparison_sum: 962892,
comparison_percent: 0,
change_sum: 50596,
change_percent: 0.05546006997728807,
change_percentage_point: 0,
difference_sum: 0,
significant: true,
difference_percent: 0

If multiplying the sum_ratio (rate) by 100 does not affect dependent variables, then it seems like it's not a problem... if it does, then we need to distinguish this mutliplication as a presentational factor...

this doesn't prevent the new ACS data from being released since it's an issue with decennial (and it's currently happening in production) but should be fixed. The rates are not an input into other variables, so fixing them shouldn't affect any other variables. oochu1 is an input into hmownrvcrt rather than the other way around.

was this pushed to staging already?

@EricaMaurer hmm it wasn't, let me see, I'll put up a PR.

NYCPlanning/labs-factfinder-api#97

I'm going to merge this because it sounds like it wouldn't affect downstream things.

It is something we want to think about though — multiply-by-100 for presentation can't always be used as inputs for other things, unless it's adjusted for.

Deploying now. @EricaMaurer please give this a look in a few minutes and see if it solves this specific problem. If it does, i'll close. Other observered side effects can be made into new issues.

Thanks!

Solved this problem! thanks!!

Great! Closing for now. Keep eyes peeled for side-effects. Thank you!!!