Add tests to make sure each source makes it to the score correctly #1878
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mattbowen-usds
changed the base branch from
main
to
emma-nechamkin/release/score-narwhal
| @@ -159,16 +158,6 @@ def extract(self) -> None: | |||
| low_memory=False, | |||
| ) | |||
|
|
|||
| # Load persistent poverty | |||
mattbowen-usds
commented
Sep 1, 2022
| def test_data_sources( | ||
| final_score_df, hud_housing_df, ejscreen_df, cdc_places_df | ||
| ): | ||
| data_sources = { |
There was a problem hiding this comment.
Like that float thing from my last PR, I'd likely never do this in production code, but in a test I am like "wheee"
| @@ -2,6 +2,9 @@ | |||
| import pytest | |||
| from data_pipeline.config import settings | |||
| from data_pipeline.score import field_names | |||
| from data_pipeline.etl.score import constants | |||
|
|
|||
| GEOID_TRACT_FIELD_NAME = field_names.GEOID_TRACT_FIELD | |||
There was a problem hiding this comment.
an extremely annoying nit: we set this constant here and then lower, we use field_names.GEOID_TRACT_FIELD anyway
| @@ -28,7 +28,6 @@ def full_percentile_column_name(self): | |||
| return self.percentile_column_name | |||
|
|
|||
|
|
|||
| ### TODO: we need to blow this out for all eight categories | |||
There was a problem hiding this comment.
thank you for the cleanup!
emma-nechamkin
approved these changes
Sep 1, 2022
| final_score_df, hud_housing_df, ejscreen_df, cdc_places_df | ||
| ): | ||
| data_sources = { | ||
| key: value for key, value in locals().items() if key != "final_score_df" |
| for col in data_source.columns | ||
| if (col != GEOID_TRACT_FIELD_NAME and col in final_score_df.columns) | ||
| ] | ||
| assert np.all(df[df.MERGE == "left_only"][final_columns].isna()) |
There was a problem hiding this comment.
this is nice. Another way to do the check is:
df[df.MERGE == "left_only"][final_columns].notna().sum() == 0
That's not necessarily better but may read a bit more easily L -> R
There was a problem hiding this comment.
(you could also do isnull().all() if you wanted it to be super easy L->R)
There was a problem hiding this comment.
isnull().all() returns a series, so it ends up needing to be isnull().all().all() which felt slightly confusing to me. Since there's a builtin all() in python, I felt okay with this being idiomatic enough.
| suffixes=(final, f"_{data_source_name}"), | ||
| how="left", | ||
| ) | ||
| data_source_columns = [ |
There was a problem hiding this comment.
I think here it might actually be easier to do the set intersection of the columns, but that's a REAL nit
There was a problem hiding this comment.
I can def at least pull the duplicated lines out, now that I look at it again.
| assert np.all(df[df.MERGE == "left_only"][final_columns].isna()) | ||
| df = df[df.MERGE == "both"] | ||
| assert ( | ||
| final_columns |
There was a problem hiding this comment.
this is neat, I would have done len(final_columns) > 0 but I like this more.
There was a problem hiding this comment.
If I'm reading this right, you're trying to make sure that at least one field from the data source's ETL shows up in the columns in the "final score", is that right?
| for final_column, data_source_column in zip( | ||
| data_source_columns, final_columns | ||
| ): | ||
| assert np.allclose( |
|
** Score Deployed! **
|
|
** Map Deployed! ** |
mattbowen-usds
force-pushed
the
mattbowen-usds/1835-source-tests
branch
from
7767ff2
to
9cd5eb6
Compare
|
** Score Deployed! **
|
|
** Score Deployed! **
|
|
** Map Deployed! ** |
|
** Map Deployed! ** |
|
** Score Deployed! **
|
mattbowen-usds
requested review from
esfoobar-usds and
vim-usds
as code owners
|
** Map Deployed! ** |
|
** Score Deployed! **
|
|
** Map Deployed! ** |
| @pytest.fixture() | ||
| def national_risk_index_df(): | ||
| return pd.read_csv( | ||
| constants.DATA_PATH / "dataset" / "national_risk_index" / "usa.csv", |
There was a problem hiding this comment.
@mattbowen-usds a design question -- for some of these, we have the class method to get the df. I assumed that loading the df fresh was helpful because it gets like, the most final possible output -- but I'm wondering if that's the right assumption here?
There was a problem hiding this comment.
I mostly just didn't want to couple to all those classes, since the overall design of the system assumes these CSVs exist. I am not like deeply committed to not using the classes --- just try to import as little of the system under test in my fixtures as I can to prevent bugs there from crashing or otherwise harming my tests.
There was a problem hiding this comment.
Maybe this is related to the question I posted and makes sense.
|
|
||
| pytestmark = pytest.mark.smoketest | ||
| GEOID_TRACT_FIELD_NAME = field_names.GEOID_TRACT_FIELD | ||
| UNMATCHED_TRACK_THRESHOLD = 1000 |
There was a problem hiding this comment.
I love this, thank you!
There was a problem hiding this comment.
(But a quick question -- why not keep them, from like an ease of use perspective, along with the fixtures?)
There was a problem hiding this comment.
I don't think I understand the question --- could you explain more of what you're thinking?
There was a problem hiding this comment.
yeah! why not just access GEOID fieldname as fixtures.GEOID...
There was a problem hiding this comment.
OHHH! The reason I defined GEOID_TRACT_FIELD_NAME = field_names.GEOID_TRACT_FIELD at the top of the file was because I knew I'd be referencing it a TON of times and honestly just wanted to save the namespace. BUT, now that I look at it again, there's a more obvious way to do that in python, so I did that in 2fac8a1
There was a problem hiding this comment.
do we actually mean TRACT not TRACK?
There was a problem hiding this comment.
Yes, I did :D
|
** Score Deployed! **
|
|
** Map Deployed! ** |
esfoobar-usds
approved these changes
Sep 6, 2022
Comment on lines
+40
to
+43
| def hud_housing_df(): | ||
| hud_housing_csv = ( | ||
| constants.DATA_PATH / "dataset" / "hud_housing" / "usa.csv" | ||
| ) |
There was a problem hiding this comment.
Idea: why re-write the paths for each dataset, when there's a class variable for it here? I'm mostly thinking about keeping it DRY?
There was a problem hiding this comment.
(this would apply to all the df's here)
There was a problem hiding this comment.
They're all instance variables and it just didn't make sense to me to import and instantiate all those classes to just get the paths. I tried to factor them up as classvars but they rely on a method so like it didn't make a ton of sense. So in this case I figured keeping the coupling lower and relying on a pretty thorough convention was worth the repetition. Open to arguments otherwise, but that as my logic.
There was a problem hiding this comment.
Makes sense, Matt, thanks! I've seen this repetition pattern in fixtures often to totally decouple things.
| @pytest.fixture() | ||
| def national_risk_index_df(): | ||
| return pd.read_csv( | ||
| constants.DATA_PATH / "dataset" / "national_risk_index" / "usa.csv", |
There was a problem hiding this comment.
Maybe this is related to the question I posted and makes sense.
mattbowen-usds
force-pushed
the
mattbowen-usds/1835-source-tests
branch
2 times, most recently
from
2fac8a1
to
2d0a4aa
Compare
mattbowen-usds
force-pushed
the
mattbowen-usds/1835-source-tests
branch
from
2d0a4aa
to
4459a77
Compare
|
** Score Deployed! **
|
|
** Score Deployed! **
|
|
** Score Deployed! **
|
|
** Score Deployed! **
|
|
** Map Deployed! ** |
|
** Map Deployed! ** |
|
** Map Deployed! ** |
|
** Map Deployed! ** |
emma-nechamkin
pushed a commit
that referenced
this pull request
Sep 7, 2022
…1878) * Remove unused persistent poverty from score (#1835) * Test a few datasets for overlap in the final score (#1835) * Add remaining data sources (#1853) * Apply code-review feedback (#1835) * Rearrange a little for readabililty (#1835) * Add tract test (#1835) * Add test for score values (#1835) * Check for unmatched source tracts (#1835) * Cleanup numeric code to plaintext (#1835) * Make import more obvious (#1835)
|
|
||
|
|
||
| @pytest.fixture() | ||
| def national_tract_df(): |
There was a problem hiding this comment.
We should add in the datasets that have been added since then, right? Like tribal_overlap.
There was a problem hiding this comment.
Yup! Added tribal overlap in #1904 since that's where it actually has output.
| key: value for key, value in locals().items() if key != "final_score_df" | ||
| } | ||
|
|
||
| for data_source_name, data_source in data_sources.items(): |
There was a problem hiding this comment.
It might be helpful to add a couple more comments about what this is doing and why. I'm having a hard time following exactly what's being tested here?
There was a problem hiding this comment.
Cool, will add a larger comment
|
|
||
| # Make sure we have NAs for any tracts in the final data that aren't | ||
| # covered in the final data | ||
| assert np.all(df[df.MERGE == "left_only"][final_columns].isna()) |
There was a problem hiding this comment.
So this asserts that for any tracts that are in final but are NOT in source, then source must have exclusively NAs for the columns from this ETL source once they show up in final?
There was a problem hiding this comment.
Yup! That's what it's trying to say at least
| df[data_source_column] | ||
| ), error_message | ||
| else: | ||
| assert np.allclose( |
|
|
||
|
|
||
| def test_all_tracts_have_scores(final_score_df): | ||
| assert not final_score_df[field_names.SCORE_N_COMMUNITIES].isna().any() |
| @@ -1,13 +1,217 @@ | |||
| import pandas as pd | |||
| import pytest | |||
| from data_pipeline.config import settings | |||
| from data_pipeline.score import field_names | |||
| from data_pipeline.score.field_names import GEOID_TRACT_FIELD | |||
| from data_pipeline.etl.score import constants | |||
|
|
|||
|
|
|||
| @pytest.fixture(scope="session") | |||
| def final_score_df(): | |||
| return pd.read_csv( | |||
| settings.APP_ROOT / "data" / "score" / "csv" / "full" / "usa.csv", | |||
There was a problem hiding this comment.
Question: will this test fail until the user has run the full score pipeline at least once on their local?
If so, that could potentially make it hard for new developers who are just "opening the box" on the CEJST.
There was a problem hiding this comment.
It will, but any tests that use it are marked as smoketest and skipped by default. I should add something to the readme tho, so I will do so.
mattbowen-usds
added a commit
that referenced
this pull request
Dec 2, 2022
* Create deploy_be_staging.yml (#1575) * Imputing income using geographic neighbors (#1559) Imputes income field with a light refactor. Needs more refactor and more tests (I spotchecked). Next ticket will check and address but a lot of "narwhal" architecture is here. * Adding HOLC indicator (#1579) Added HOLC indicator (Historic Redlining Score) from NCRC work; included 3.25 cutoff and low income as part of the housing burden category. * Update backend for Puerto Rico (#1686) * Update PR threshold count to 10 We now show 10 indicators for PR. See the discussion on the github issue for more info: #1621 * Do not use linguistic iso for Puerto Rico Closes 1350. Co-authored-by: Shelby Switzer <[email protected]> * updating * Do not drop Guam and USVI from ETL (#1681) * Remove code that drops Guam and USVI from ETL * Add back code for dropping rows by FIPS code We may want this functionality, so let's keep it and just make the constant currently be an empty array. Co-authored-by: Shelby Switzer <[email protected]> * Emma nechamkin/holc patch (#1742) Removing HOLC calculation from score narwhal. * updating ejscreen data, try two (#1747) * Rescaling linguistic isolation (#1750) Rescales linguistic isolation to drop puerto rico * adds UST indicator (#1786) adds leaky underground storage tanks * Changing LHE in tiles to a boolean (#1767) also includes merging / clean up of the release * added indoor plumbing to chas * added indoor plumbing to score housing burden * added indoor plumbing to score housing burden * first run through * Refactor DOE Energy Burden and COI to use YAML (#1796) * added tribalId for Supplemental dataset (#1804) * Setting zoom levels for tribal map (#1810) * NRI dataset and initial score YAML configuration (#1534) * update be staging gha * NRI dataset and initial score YAML configuration * checkpoint * adding data checks for release branch * passing tests * adding INPUT_EXTRACTED_FILE_NAME to base class * lint * columns to keep and tests * update be staging gha * checkpoint * update be staging gha * NRI dataset and initial score YAML configuration * checkpoint * adding data checks for release branch * passing tests * adding INPUT_EXTRACTED_FILE_NAME to base class * lint * columns to keep and tests * checkpoint * PR Review * renoving source url * tests * stop execution of ETL if there's a YAML schema issue * update be staging gha * adding source url as class var again * clean up * force cache bust * gha cache bust * dynamically set score vars from YAML * docsctrings * removing last updated year - optional reverse percentile * passing tests * sort order * column ordening * PR review * class level vars * Updating DatasetsConfig * fix pylint errors * moving metadata hint back to code Co-authored-by: lucasmbrown-usds <[email protected]> * Correct copy typo (#1809) * Add basic test suite for COI (#1518) * Update COI to use new yaml (#1518) * Add tests for DOE energy budren (1518 * Add dataset config for energy budren (1518) * Refactor ETL to use datasets.yml (#1518) * Add fake GEOIDs to COI tests (#1518) * Refactor _setup_etl_instance_and_run_extract to base (#1518) For the three classes we've done so far, a generic _setup_etl_instance_and_run_extract will work fine, for the moment we can reuse the same setup method until we decide future classes need more flexibility --- but they can also always subclass so... * Add output-path tests (#1518) * Update YAML to match constant (#1518) * Don't blindly set float format (#1518) * Add defaults for extract (#1518) * Run YAML load on all subclasses (#1518) * Update description fields (#1518) * Update YAML per final format (#1518) * Update fixture tract IDs (#1518) * Update base class refactor (#1518) Now that NRI is final I needed to make a small number of updates to my refactored code. * Remove old comment (#1518) * Fix type signature and return (#1518) * Update per code review (#1518) Co-authored-by: Jorge Escobar <[email protected]> Co-authored-by: lucasmbrown-usds <[email protected]> Co-authored-by: Vim <[email protected]> * Update etl_score_geo.py Yikes! Fixing merge messup! * Create deploy_be_staging.yml (#1575) * Imputing income using geographic neighbors (#1559) Imputes income field with a light refactor. Needs more refactor and more tests (I spotchecked). Next ticket will check and address but a lot of "narwhal" architecture is here. * Adding HOLC indicator (#1579) Added HOLC indicator (Historic Redlining Score) from NCRC work; included 3.25 cutoff and low income as part of the housing burden category. * Update backend for Puerto Rico (#1686) * Update PR threshold count to 10 We now show 10 indicators for PR. See the discussion on the github issue for more info: #1621 * Do not use linguistic iso for Puerto Rico Closes 1350. Co-authored-by: Shelby Switzer <[email protected]> * updating * Do not drop Guam and USVI from ETL (#1681) * Remove code that drops Guam and USVI from ETL * Add back code for dropping rows by FIPS code We may want this functionality, so let's keep it and just make the constant currently be an empty array. Co-authored-by: Shelby Switzer <[email protected]> * Emma nechamkin/holc patch (#1742) Removing HOLC calculation from score narwhal. * updating ejscreen data, try two (#1747) * Rescaling linguistic isolation (#1750) Rescales linguistic isolation to drop puerto rico * adds UST indicator (#1786) adds leaky underground storage tanks * Changing LHE in tiles to a boolean (#1767) also includes merging / clean up of the release * added indoor plumbing to chas * added indoor plumbing to score housing burden * added indoor plumbing to score housing burden * first run through * Refactor DOE Energy Burden and COI to use YAML (#1796) * added tribalId for Supplemental dataset (#1804) * Setting zoom levels for tribal map (#1810) * NRI dataset and initial score YAML configuration (#1534) * update be staging gha * NRI dataset and initial score YAML configuration * checkpoint * adding data checks for release branch * passing tests * adding INPUT_EXTRACTED_FILE_NAME to base class * lint * columns to keep and tests * update be staging gha * checkpoint * update be staging gha * NRI dataset and initial score YAML configuration * checkpoint * adding data checks for release branch * passing tests * adding INPUT_EXTRACTED_FILE_NAME to base class * lint * columns to keep and tests * checkpoint * PR Review * renoving source url * tests * stop execution of ETL if there's a YAML schema issue * update be staging gha * adding source url as class var again * clean up * force cache bust * gha cache bust * dynamically set score vars from YAML * docsctrings * removing last updated year - optional reverse percentile * passing tests * sort order * column ordening * PR review * class level vars * Updating DatasetsConfig * fix pylint errors * moving metadata hint back to code Co-authored-by: lucasmbrown-usds <[email protected]> * Correct copy typo (#1809) * Add basic test suite for COI (#1518) * Update COI to use new yaml (#1518) * Add tests for DOE energy budren (1518 * Add dataset config for energy budren (1518) * Refactor ETL to use datasets.yml (#1518) * Add fake GEOIDs to COI tests (#1518) * Refactor _setup_etl_instance_and_run_extract to base (#1518) For the three classes we've done so far, a generic _setup_etl_instance_and_run_extract will work fine, for the moment we can reuse the same setup method until we decide future classes need more flexibility --- but they can also always subclass so... * Add output-path tests (#1518) * Update YAML to match constant (#1518) * Don't blindly set float format (#1518) * Add defaults for extract (#1518) * Run YAML load on all subclasses (#1518) * Update description fields (#1518) * Update YAML per final format (#1518) * Update fixture tract IDs (#1518) * Update base class refactor (#1518) Now that NRI is final I needed to make a small number of updates to my refactored code. * Remove old comment (#1518) * Fix type signature and return (#1518) * Update per code review (#1518) Co-authored-by: Jorge Escobar <[email protected]> Co-authored-by: lucasmbrown-usds <[email protected]> Co-authored-by: Vim <[email protected]> * Update etl_score_geo.py Yikes! Fixing merge messup! * updated to fix linting errors (#1818) Cleans and updates base branch * Adding back MapComparison video * Add FUDS ETL (#1817) * Add spatial join method (#1871) Since we'll need to figure out the tracts for a large number of points in future tickets, add a utility to handle grabbing the tract geometries and adding tract data to a point dataset. * Add FUDS, also jupyter lab (#1871) * Add YAML configs for FUDS (#1871) * Allow input geoid to be optional (#1871) * Add FUDS ETL, tests, test-datae noteobook (#1871) This adds the ETL class for Formerly Used Defense Sites (FUDS). This is different from most other ETLs since these FUDS are not provided by tract, but instead by geographic point, so we need to assign FUDS to tracts and then do calculations from there. * Floats -> Ints, as I intended (#1871) * Floats -> Ints, as I intended (#1871) * Formatting fixes (#1871) * Add test false positive GEOIDs (#1871) * Add gdal binaries (#1871) * Refactor pandas code to be more idiomatic (#1871) Per Emma, the more pandas-y way of doing my counts is using np.where to add the values i need, then groupby and size. It is definitely more compact, and also I think more correct! * Update configs per Emma suggestions (#1871) * Type fixed! (#1871) * Remove spurious import from vscode (#1871) * Snapshot update after changing col name (#1871) * Move up GDAL (#1871) * Adjust geojson strategy (#1871) * Try running census separately first (#1871) * Fix import order (#1871) * Cleanup cache strategy (#1871) * Download census data from S3 instead of re-calculating (#1871) * Clarify pandas code per Emma (#1871) * Disable markdown check for link * Adding DOT composite to travel score (#1820) This adds the DOT dataset to the ETL and to the score. Note that currently we take a percentile of an average of percentiles. * Adding first street foundation data (#1823) Adding FSF flood and wildfire risk datasets to the score. * first run -- adding NCLD data to the ETL, but not yet to the score * Add abandoned mine lands data (#1824) * Add notebook to generate test data (#1780) * Add Abandoned Mine Land data (#1780) Using a similar structure but simpler apporach compared to FUDs, add an indicator for whether a tract has an abandonded mine. * Adding some detail to dataset readmes Just a thought! * Apply feedback from revieiw (#1780) * Fixup bad string that broke test (#1780) * Update a string that I should have renamed (#1780) * Reduce number of threads to reduce memory pressure (#1780) * Try not running geo data (#1780) * Run the high-memory sets separately (#1780) * Actually deduplicate (#1780) * Add flag for memory intensive ETLs (#1780) * Document new flag for datasets (#1780) * Add flag for new datasets fro rebase (#1780) Co-authored-by: Emma Nechamkin <[email protected]> * Adding NLCD data (#1826) Adding NLCD's natural space indicator end to end to the score. * Add donut hole calculation to score (#1828) Adds adjacency index to the pipeline. Requires thorough QA * Adding eamlis and fuds data to legacy pollution in score (#1832) Update to add EAMLIS and FUDS data to score * Update to use new FSF files (#1838) backend is partially done! * Quick fix to kitchen or plumbing indicator Yikes! I think I messed something up and dropped the pctile field suffix from when the KP score gets calculated. Fixing right quick. * Fast flag update (#1844) Added additional flags for the front end based on our conversation in stand up this morning. * Tiles fix (#1845) Fixes score-geo and adds flags * Update etl_score_geo.py * Issue 1827: Add demographics to tiles and download files (#1833) * Adding demographics for use in sidebar and download files * Updates backend constants to N (#1854) * updated to show T/F/null vs T/F for AML and FUDS (#1866) * fix markdown * just testing that the boolean is preserved on gha * checking drop tracts works * OOPS! Old changes persisted * adding a check to the agvalue calculation for nri * updated with error messages * updated error message * tuple type * Score tests (#1847) * update Python version on README; tuple typing fix * Alaska tribal points fix (#1821) * Bump mistune from 0.8.4 to 2.0.3 in /data/data-pipeline (#1777) Bumps [mistune](https://github.com/lepture/mistune) from 0.8.4 to 2.0.3. - [Release notes](https://github.com/lepture/mistune/releases) - [Changelog](https://github.com/lepture/mistune/blob/master/docs/changes.rst) - [Commits](lepture/mistune@v0.8.4...v2.0.3) --- updated-dependencies: - dependency-name: mistune dependency-type: indirect ... Signed-off-by: dependabot[bot] <[email protected]> Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * poetry update * initial pass of score tests * add threshold tests * added ses threshold (not donut, not island) * testing suite -- stopping for the day * added test for lead proxy indicator * Refactor score tests to make them less verbose and more direct (#1865) * Cleanup tests slightly before refactor (#1846) * Refactor score calculations tests * Feedback from review * Refactor output tests like calculatoin tests (#1846) (#1870) * Reorganize files (#1846) * Switch from lru_cache to fixture scorpes (#1846) * Add tests for all factors (#1846) * Mark smoketests and run as part of be deply (#1846) * Update renamed var (#1846) * Switch from named tuple to dataclass (#1846) This is annoying, but pylint in python3.8 was crashing parsing the named tuple. We weren't using any namedtuple-specific features, so I made the type a dataclass just to get pylint to behave. * Add default timout to requests (#1846) * Fix type (#1846) * Fix merge mistake on poetry.lock (#1846) Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: Jorge Escobar <[email protected]> Co-authored-by: Jorge Escobar <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Matt Bowen <[email protected]> Co-authored-by: matt bowen <[email protected]> * just testing that the boolean is preserved on gha (#1867) * updated with hopefully a fix; coercing aml, fuds, hrs to booleans for the raw value to preserve null character. * Adding tests to ensure proper calculations (#1871) * just testing that the boolean is preserved on gha * checking drop tracts works * adding a check to the agvalue calculation for nri * updated with error messages * tribal tiles fix (#1874) * Alaska tribal points fix (#1821) * tribal tiles fix * disabling child opportunity * lint * removing COI * removing commented out code * Pipeline tile tests (#1864) * temp update * updating with fips check * adding check on pfs * updating with pfs test * Update test_tiles_smoketests.py * Fix lint errors (#1848) * Add column names test (#1848) * Mark tests as smoketests (#1848) * Move to other score-related tests (#1848) * Recast Total threshold criteria exceeded to int (#1848) In writing tests to verify the output of the tiles csv matches the final score CSV, I noticed TC/Total threshold criteria exceeded was getting cast from an int64 to a float64 in the process of PostScoreETL. I tracked it down to the line where we merge the score dataframe with constants.DATA_CENSUS_CSV_FILE_PATH --- there where > 100 tracts in the national census CSV that don't exist in the score, so those ended up with a Total threshhold count of np.nan, which is a float, and thereby cast those columns to float. For the moment I just cast it back. * No need for low memeory (#1848) * Add additional tests of tiles.csv (#1848) * Drop pre-2010 rows before computing score (#1848) Note this is probably NOT the optimal place for this change; it might make more sense for each source to filter its own tracts down to the acceptable tract list. However, that would be a pretty invasive change, where this is central and plenty of other things are happening in score transform that could be moved to sources, so for today, here's where the change will live. * Fix typo (#1848) * Switch from filter to inner join (#1848) * Remove no-op lines from tiles (#1848) * Apply feedback from review, linter (#1848) * Check the values oeverything in the frame (#1848) * Refactor checker class (#1848) * Add test for state names (#1848) * cleanup from reviewing my own code (#1848) * Fix lint error (#1858) * Apply Emma's feedback from review (#1848) * Remove refs to national_df (#1848) * Account for new, fake nullable bools in tiles (#1848) To handle a geojson limitation, Emma converted some nullable boolean colunms to float64 in the tiles export with the values {0.0, 1.0, nan}, giving us the same expressiveness. Sadly, this broke my assumption that all columns between the score and tiles csvs would have the same dtypes, so I need to account for these new, fake bools in my test. * Use equals instead of my worse version (#1848) * Missed a spot where we called _create_score_data (#1848) * Update per safety (#1848) Co-authored-by: matt bowen <[email protected]> * Add tests to make sure each source makes it to the score correctly (#1878) * Remove unused persistent poverty from score (#1835) * Test a few datasets for overlap in the final score (#1835) * Add remaining data sources (#1853) * Apply code-review feedback (#1835) * Rearrange a little for readabililty (#1835) * Add tract test (#1835) * Add test for score values (#1835) * Check for unmatched source tracts (#1835) * Cleanup numeric code to plaintext (#1835) * Make import more obvious (#1835) * Updating traffic barriers to include low pop threshold (#1889) Changing the traffic barriers to only be included for places with recorded population * Remove no land tracts from map (#1894) remove from map * Issue 1831: missing life expectancy data from Maine and Wisconsin (#1887) * Fixing missing states and adding tests for states to all classes * Removing low pop tracts from FEMA population loss (#1898) dropping 0 population from FEMA * 1831 Follow up (#1902) This code causes no functional change to the code. It does two things: 1. Uses difference instead of - to improve code style for working with sets. 2. Removes the line EXPECTED_MISSING_STATES = ["02", "15"], which is now redundant because of the line I added (in a previous pull request) of ALASKA_AND_HAWAII_EXPECTED_IN_DATA = False. * Add tests for all non-census sources (#1899) * Refactor CDC life-expectancy (1554) * Update to new tract list (#1554) * Adjust for tests (#1848) * Add tests for cdc_places (#1848) * Add EJScreen tests (#1848) * Add tests for HUD housing (#1848) * Add tests for GeoCorr (#1848) * Add persistent poverty tests (#1848) * Update for sources without zips, for new validation (#1848) * Update tests for new multi-CSV but (#1848) Lucas updated the CDC life expectancy data to handle a bug where two states are missing from the US Overall download. Since virtually none of our other ETL classes download multiple CSVs directly like this, it required a pretty invasive new mocking strategy. * Add basic tests for nature deprived (#1848) * Add wildfire tests (#1848) * Add flood risk tests (#1848) * Add DOT travel tests (#1848) * Add historic redlining tests (#1848) * Add tests for ME and WI (#1848) * Update now that validation exists (#1848) * Adjust for validation (#1848) * Add health insurance back to cdc places (#1848) Ooops * Update tests with new field (#1848) * Test for blank tract removal (#1848) * Add tracts for clipping behavior * Test clipping and zfill behavior (#1848) * Fix bad test assumption (#1848) * Simplify class, add test for tract padding (#1848) * Fix percentage inversion, update tests (#1848) Looking through the transformations, I noticed that we were subtracting a percentage that is usually between 0-100 from 1 instead of 100, and so were endind up with some surprising results. Confirmed with lucasmbrown-usds * Add note about first street data (#1848) * Issue 1900: Tribal overlap with Census tracts (#1903) * working notebook * updating notebook * wip * fixing broken tests * adding tribal overlap files * WIP * WIP * WIP, calculated count and names * working * partial cleanup * partial cleanup * updating field names * fixing bug * removing pyogrio * removing unused imports * updating test fixtures to be more realistic * cleaning up notebook * fixing black * fixing flake8 errors * adding tox instructions * updating etl_score * suppressing warning * Use projected CRSes, ignore geom types (#1900) I looked into this a bit, and in general the geometry type mismatch changes very little about the calculation; we have a mix of multipolygons and polygons. The fastest thing to do is just not keep geom type; I did some runs with it set to both True and False, and they're the same within 9 digits of precision. Logically we just want to overlaps, regardless of how the actual geometries are encoded between the frames, so we can in this case ignore the geom types and feel OKAY. I also moved to projected CRSes, since we are actually trying to do area calculations and so like, we should. Again, the change is small in magnitude but logically more sound. * Readd CDC dataset config (#1900) * adding comments to fips code * delete unnecessary loggers Co-authored-by: matt bowen <[email protected]> * Improve score test documentation based on Lucas's feedback (#1835) (#1914) * Better document base on Lucas's feedback (#1835) * Fix typo (#1835) * Add test to verify GEOJSON matches tiles (#1835) * Remove NOOP line (#1835) * Move GEOJSON generation up for new smoketest (#1835) * Fixup code format (#1835) * Update readme for new somketest (#1835) * Cleanup source tests (#1912) * Move test to base for broader coverage (#1848) * Remove duplicate line (#1848) * FUDS needed an extra mock (#1848) * Add tribal count notebook (#1917) (#1919) * Add tribal count notebook (#1917) * test without caching * added comment Co-authored-by: lucasmbrown-usds <[email protected]> * Add tribal overlap to downloads (#1907) * Add tribal data to downloads (#1904) * Update test pickle with current cols (#1904) * Remove text of tribe names from GeoJSON (#1904) * Update test data (#1904) * Add tribal overlap to smoketests (#1904) * Issue 1910: Do not impute income for 0 population tracts (#1918) * should be working, has unnecessary loggers * removing loggers and cleaning up * updating ejscreen tests * adding tests and responding to PR feedback * fixing broken smoke test * delete smoketest docs * updating click * updating click * Bump just jupyterlab (#1930) * Fixing link checker (#1929) * Update deps safety says are vulnerable (#1937) (#1938) Co-authored-by: matt bowen <[email protected]> * Add demos for island areas (#1932) * Backfill population in island areas (#1882) * Update smoketest to account for backfills (#1882) As I wrote in the commend: We backfill island areas with data from the 2010 census, so if THOSE tracts have data beyond the data source, that's to be expected and is fine to pass. If some other state or territory does though, this should fail This ends up being a nice way of documenting that behavior i guess! * Fixup lint issues (#1882) * Add in race demos to 2010 census pull (#1851) * Add backfill data to score (#1851) * Change column name (#1851) * Fill demos after the score (#1851) * Add income back, adjust test (#1882) * Apply code-review feedback (#1851) * Add test for island area backfill (#1851) * Fix bad rename (#1851) * Reorder download fields, add plumbing back (#1942) * Add back lack of plumbing fields (#1920) * Reorder fields for excel (#1921) * Reorder excel fields (#1921) * Fix formating, lint errors, pickes (#1921) * Add missing plumbing col, fix order again (#1921) * Update that pickle (#1921) * refactoring tribal (#1960) * updated with scoring comparison * updated for narhwal -- leaving commented code in for now * pydantic upgrade * produce a string for the front end to ingest (#1963) * wip * i believe this works -- let's see the pipeline * updated fixtures * Adding ADJLI_ET (#1976) * updated tile data * ensuring adjli_et in * Add back income percentile (#1977) * Add missing field to download (#1964) * Remove pydantic since it's unused (#1964) * Add percentile to CSV (#1964) * Update downloadable pickle (#1964) * Issue 105: Configure and run `black` and other pre-commit hooks (clean branch) (#1962) * Configure and run `black` and other pre-commit hooks Co-authored-by: matt bowen <[email protected]> * Removing fixed python version for black (#1985) * Fixup TA_COUNT and TA_PERC (#1991) * Change TA_PERC, change TA_COUNT (#1988, #1989) - Make TA_PERC_STR back into a nullable float following the rules requestsed in #1989 - Move TA_COUNT to be TA_COUNT_AK, also add a null TA_COUNT_C for CONUS that we can fill in later. * Fix typo comment (#1988) * Issue 1992: Do not impute income for null population tracts (#1993) * Hotfix for DOT data source DNS issue (#1999) * Make tribal overlap set score N (#2004) * Add "Is a Tribal DAC" field (#1998) * Add tribal DACs to score N final (#1998) * Add new fields to downloads (#1998) * Make a int a float (#1998) * Update field names, apply feedback (#1998) * Add assertions around codebook (#2014) * Add assertion around codebook (#1505) * Assert csv and excel have same cols (#1505) * Remove suffixes from tribal lands (#1974) (#2008) * Data source location (#2015) * data source location * toml * cdc_places * cdc_svi_index * url updates * child oppy and dot travel * up to hud_recap * completed ticket * cache bust * hud_recap * us_army_fuds * Remove vars the frontend doesn't use (#2020) (#2022) I did a pretty rough and simple analysis of the variables we put in the tiles and grepped the frontend code to see if (1) they're ever accessed and (2) if they're used, even if they're read once. I removed everything I noticed was not accessed. * Disable file size limits on tiles (#2031) * Disable file size limits on tiles * Remove print debugs I know. * Update file name pattern (#2037) (#2038) * Update file name pattern (#2037) * Remove ETL from generation (2037) I looked more carefully, and this ETL step isn't used in the score, so there's no need to run it every time. Per previous steps, I removed it from constants so the code is there it won't run by default. * Round ALL the float fields for the tiles (#2040) * Round ALL the float fields for the tiles (#2033) * Floor in a simpler way (#2033) Emma pointed out that all teh stuff we're doing in floor_series is probably unnecessary for this case, so just use the built-in floor. * Update pickle I missed (#2033) * Clean commit of just aggregate burden notebook (#1819) added a burden notebook * Update the dockerfile (#2045) * Update so the image builds (#2026) * Fix bad dict (2026) * Rename census tract field in downloads (#2068) * Change tract ID field name (2060) * Update lockfile (#2061) * Bump safety, jupyter, wheel (#2061) * DOn't depend directly on wheel (2061) * Bring narwhal reqs in line with main * Update tribal area counts (#2071) * Rename tribal area field (2062) * Add missing file (#2062) * Add checks to create version (#2047) (#2052) * Fix failing safety (#2114) * Ignore vuln that doesn't affect us 2113 https://nvd.nist.gov/vuln/detail/CVE-2022-42969 landed recently and there's no fix in py (which is maintenance mode). From my analysis, that CVE cannot hurt us (famous last words), so we'll ignore the vuln for now. * 2113 Update our gdal ppa * that didn't work (2113) * Don't add the PPA, the package exists (#2113) * Fix type (#2113) * Force an update of wheel 2113 * Also remove PPA line from create-score-versions * Drop 3.8 because of wheel 2113 * Put back 3.8, use newer actions * Try another way of upgrading wheel 2113 * Upgrade wheel in tox too 2113 * Typo fix 2113 Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: Emma Nechamkin <[email protected]> Co-authored-by: Shelby Switzer <[email protected]> Co-authored-by: Shelby Switzer <[email protected]> Co-authored-by: Emma Nechamkin <[email protected]> Co-authored-by: Matt Bowen <[email protected]> Co-authored-by: Jorge Escobar <[email protected]> Co-authored-by: lucasmbrown-usds <[email protected]> Co-authored-by: Jorge Escobar <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: matt bowen <[email protected]> Co-authored-by: matt bowen <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Closes #1835