Remove OpenFisca-UK fork dependency + other changes #139
Conversation
| - name: Test baseline systems | ||
| run: pytest tests/code | ||
|
|
||
| - name: Test baseline systems |
There was a problem hiding this comment.
should this have a different name from above?
There was a problem hiding this comment.
Yep, changed
| @@ -0,0 +1,380 @@ | |||
| ,LAD20CD,LAD20NM | |||
There was a problem hiding this comment.
Nope, removed
| @@ -0,0 +1,380 @@ | |||
| ,LAD20CD,LAD20NM | |||
There was a problem hiding this comment.
could you add a README.md to this folder specifying the data source?
| @@ -341,9 +341,6 @@ def apply(self): | |||
| "Unable to compute derivative - target variable must be from a group of or the same as the source variable" | |||
| ) | |||
|
|
|||
| def mtr(self, *args, **kwargs): | |||
There was a problem hiding this comment.
So users should now use deriv to calculate MTRs?
There was a problem hiding this comment.
Yep, and this removes ambiguity. The current documentation doesn't detail this, I've added an issue at #141 .
| reference = "Child Tax Credit Regulations 2002 s. 8" | ||
|
|
||
| def formula(person, period, parameters): | ||
| benefit = parameters(period).benefit | ||
| threshold_safety_gap = 10 | ||
| THRESHOLD_SAFETY_GAP = 10 * WEEKS_IN_YEAR |
There was a problem hiding this comment.
what is this? should 10 be a policy parameter?
There was a problem hiding this comment.
This should really be documented, but from memory it's because there are some small variations around the disability benefits paid, probably because of weeklyisation/irregular payments. But where we're imputing the basic/higher levels of benefits based on their amounts, we err on the side of being generous (£10/week is less than the difference between the levels).
| add(benunit, period, ["working_tax_credit"], options=[DIVIDE]) | ||
| + add(benunit, period, ["ESA_income"], options=[ADD]) | ||
| + aggr(benunit, period, QUAL_PERSONAL_BENEFITS, options=[ADD]) | ||
| add(benunit, period, ["working_tax_credit"]) |
There was a problem hiding this comment.
How does it know to divide this one and add the next one? In general since this PR removes a bunch of options= args could you share how this works?
There was a problem hiding this comment.
To recap here, it's because by making everything yearly, we've shifted from doing the adding/dividing time periods in this specific variables to doing the same operations within working_tax_credit, ESA_income and the others.
| @@ -1,12 +0,0 @@ | |||
| - name: Poverty gap for single person | |||
There was a problem hiding this comment.
Hasn't been deleted, just moved (as has disability.yaml).
|
@jpycroft, @jdebacker, @rickecon FYI this also relaxes the requirements from == to >=. |
Moderately-sized update to OpenFisca-UK (with a version change from 0.3.0 to 0.4.0):
Back on the official OpenFisca-Core
This PR removes the dependency for openfisca-uk to run off the fork of OpenFisca-Core. What did it cost? All variables are now yearly. There are pros and cons to this, but after considering them, I think the pros really outweigh with very little contest:
Reform tests and postcode handling
In preparation for using OpenFisca-UK as a tax-benefit web calculator, I added functionality to take a household's postcode as input (calculating LHA/etc downstream). I've also added the first reform test as a demonstration.
Code variable naming convention testing
I added a test that checks variable names against their return types, using the convention that was mostly (but inconsistently) followed:
@MaxGhenis