Merge pull request #45 from ihmeuw/develop

Release v0.3.2
ihmeuw · Apr 10, 2023 · d2a4109 · d2a4109
2 parents a30dab8 + 1bf63cd
commit d2a4109
Show file tree

Hide file tree

Showing 21 changed files with 19 additions and 9 deletions.
diff --git a/CHANGELOG.rst b/CHANGELOG.rst
@@ -1,3 +1,7 @@
+**0.3.2 - 04/10/23**
+
+ - Update sample datasets
+
 **0.3.1 - 04/10/23**
 
  - Build docs to readthedocs

diff --git a/MANIFEST.in b/MANIFEST.in
@@ -8,5 +8,5 @@ include README.rst
 recursive-include docs *
 prune docs/_build
 
-recursive-include src/pseudopeople *.py *.yaml *.csv *.hdf
+recursive-include src/pseudopeople *.py *.yaml *.csv *.parquet
 recursive-include tests *.py *txt *.yaml
diff --git a/src/pseudopeople/__about__.py b/src/pseudopeople/__about__.py
@@ -13,7 +13,7 @@
 __summary__ = "pseudopeople is package which adds noise to simulated census-scale data using standard scientific Python tools."
 __uri__ = "https://github.com/ihmeuw/pseudopeople"
 
-__version__ = "0.3.1"
+__version__ = "0.3.2"
 
 __author__ = "The pseudopeople developers"
 __email__ = "[email protected]"

diff --git a/src/pseudopeople/constants/paths.py b/src/pseudopeople/constants/paths.py
@@ -9,9 +9,11 @@
 QWERTY_ERRORS = DATA_ROOT / "qwerty_errors.yaml"
 
 SAMPLE_DATA_ROOT = DATA_ROOT / "sample_forms"
-SAMPLE_DECENNIAL_CENSUS = SAMPLE_DATA_ROOT / "decennial_census_observer.hdf"
-SAMPLE_TAXES_W2_AND_1099 = SAMPLE_DATA_ROOT / "tax_w2_observer.hdf"
-SAMPLE_AMERICAN_COMMUNITIES_SURVEY = SAMPLE_DATA_ROOT / "household_survey_observer_acs.hdf"
-SAMPLE_CURRENT_POPULATION_SURVEY = SAMPLE_DATA_ROOT / "household_survey_observer_cps.hdf"
-SAMPLE_SOCIAL_SECURITY = SAMPLE_DATA_ROOT / "social_security_observer.hdf"
-SAMPLE_WOMEN_INFANTS_AND_CHILDREN = SAMPLE_DATA_ROOT / "wic_observer.hdf"
+SAMPLE_DECENNIAL_CENSUS = SAMPLE_DATA_ROOT / "decennial_census_observer.parquet"
+SAMPLE_TAXES_W2_AND_1099 = SAMPLE_DATA_ROOT / "tax_w2_observer.parquet"
+SAMPLE_AMERICAN_COMMUNITIES_SURVEY = (
+    SAMPLE_DATA_ROOT / "household_survey_observer_acs.parquet"
+)
+SAMPLE_CURRENT_POPULATION_SURVEY = SAMPLE_DATA_ROOT / "household_survey_observer_cps.parquet"
+SAMPLE_SOCIAL_SECURITY = SAMPLE_DATA_ROOT / "social_security_observer.parquet"
+SAMPLE_WOMEN_INFANTS_AND_CHILDREN = SAMPLE_DATA_ROOT / "wic_observer.parquet"
diff --git a/src/pseudopeople/data/sample_forms/decennial_census_observer.hdf b/src/pseudopeople/data/sample_forms/decennial_census_observer.hdf
diff --git a/src/pseudopeople/data/sample_forms/decennial_census_observer.parquet b/src/pseudopeople/data/sample_forms/decennial_census_observer.parquet
diff --git a/src/pseudopeople/data/sample_forms/household_survey_observer_acs.hdf b/src/pseudopeople/data/sample_forms/household_survey_observer_acs.hdf
diff --git a/src/pseudopeople/data/sample_forms/household_survey_observer_acs.parquet b/src/pseudopeople/data/sample_forms/household_survey_observer_acs.parquet
diff --git a/src/pseudopeople/data/sample_forms/household_survey_observer_cps.hdf b/src/pseudopeople/data/sample_forms/household_survey_observer_cps.hdf
diff --git a/src/pseudopeople/data/sample_forms/household_survey_observer_cps.parquet b/src/pseudopeople/data/sample_forms/household_survey_observer_cps.parquet
diff --git a/src/pseudopeople/data/sample_forms/social_security_observer.hdf b/src/pseudopeople/data/sample_forms/social_security_observer.hdf
diff --git a/src/pseudopeople/data/sample_forms/social_security_observer.parquet b/src/pseudopeople/data/sample_forms/social_security_observer.parquet
diff --git a/src/pseudopeople/data/sample_forms/tax_1040_observer.hdf b/src/pseudopeople/data/sample_forms/tax_1040_observer.hdf
diff --git a/src/pseudopeople/data/sample_forms/tax_1040_observer.parquet b/src/pseudopeople/data/sample_forms/tax_1040_observer.parquet
diff --git a/src/pseudopeople/data/sample_forms/tax_dependents_observer.hdf b/src/pseudopeople/data/sample_forms/tax_dependents_observer.hdf
diff --git a/src/pseudopeople/data/sample_forms/tax_dependents_observer.parquet b/src/pseudopeople/data/sample_forms/tax_dependents_observer.parquet
diff --git a/src/pseudopeople/data/sample_forms/tax_w2_observer.hdf b/src/pseudopeople/data/sample_forms/tax_w2_observer.hdf
diff --git a/src/pseudopeople/data/sample_forms/tax_w2_observer.parquet b/src/pseudopeople/data/sample_forms/tax_w2_observer.parquet
diff --git a/src/pseudopeople/data/sample_forms/wic_observer.hdf b/src/pseudopeople/data/sample_forms/wic_observer.hdf
diff --git a/src/pseudopeople/data/sample_forms/wic_observer.parquet b/src/pseudopeople/data/sample_forms/wic_observer.parquet
diff --git a/tests/integration/test_interface.py b/tests/integration/test_interface.py
@@ -34,7 +34,11 @@
     ],
 )
 def test_generate_form(data_path: Union[Path, str], noising_function: Callable):
-    data = pd.DataFrame(pd.read_hdf(data_path))
+    data_path = Path(data_path)
+    if data_path.suffix == ".hdf":
+        data = pd.read_hdf(data_path)
+    elif data_path.suffix == ".parquet":
+        data = pd.read_parquet(data_path)
 
     noised_data = noising_function(source=data.copy(), seed=0)
     noised_data_same_seed = noising_function(source=data.copy(), seed=0)