francois/HowBadIsMyBatch
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refining SymptomsByBatchcodesTableFactoryTest

Browse Source
This commit is contained in:
frankknoll
2023-01-20 01:07:17 +01:00
parent 380721392f
commit 00da43dd5d
2 changed files with 14 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
src/SymptomsByBatchcodesTableFactory.py
View File

@@ -5,12 +5,20 @@ class SymptomsByBatchcodesTableFactory:
@staticmethod @staticmethod
def createSymptomsByBatchcodesTable(VAERSVAX, VAERSSYMPTOMS): def createSymptomsByBatchcodesTable(VAERSVAX, VAERSSYMPTOMS):
index_columns = ['VAX_LOT1', 'VAX_LOT2'] index_columns = SymptomsByBatchcodesTableFactory._getIndexColumns(VAERSVAX)
return pd.merge( return pd.merge(
SymptomsByBatchcodesTableFactory._get_VAERSVAX_WITH_VAX_LOTS(VAERSVAX, index_columns), SymptomsByBatchcodesTableFactory._get_VAERSVAX_WITH_VAX_LOTS(VAERSVAX, index_columns),
SymptomsByBatchcodesTableFactory._getSymptomsTable(VAERSSYMPTOMS), SymptomsByBatchcodesTableFactory._getSymptomsTable(VAERSSYMPTOMS),
on = 'VAERS_ID').set_index(index_columns)[['SYMPTOMS']] on = 'VAERS_ID').set_index(index_columns)[['SYMPTOMS']]
@staticmethod
def _getIndexColumns(VAERSVAX):
return [f"VAX_LOT{num}" for num in range(1, SymptomsByBatchcodesTableFactory._getMaxNumShots(VAERSVAX) + 1)]
@staticmethod
def _getMaxNumShots(VAERSVAX):
return VAERSVAX.index.value_counts().iloc[0]
@staticmethod @staticmethod
def _get_VAERSVAX_WITH_VAX_LOTS(VAERSVAX, index_columns): def _get_VAERSVAX_WITH_VAX_LOTS(VAERSVAX, index_columns):
return pd.concat( return pd.concat(
@@ -21,7 +29,7 @@ class SymptomsByBatchcodesTableFactory:
def _getVaxLotsTable(VAERSVAX, index_columns): def _getVaxLotsTable(VAERSVAX, index_columns):
VAX_LOT_LIST_Table = VAERSVAX.groupby("VAERS_ID").agg(VAX_LOT_LIST = pd.NamedAgg(column = 'VAX_LOT', aggfunc = list)) VAX_LOT_LIST_Table = VAERSVAX.groupby("VAERS_ID").agg(VAX_LOT_LIST = pd.NamedAgg(column = 'VAX_LOT', aggfunc = list))
return pd.DataFrame( return pd.DataFrame(
[fill(VAX_LOTS, 2, str(np.nan)) for VAX_LOTS in VAX_LOT_LIST_Table['VAX_LOT_LIST'].tolist()], [fill(VAX_LOTS, len(index_columns), str(np.nan)) for VAX_LOTS in VAX_LOT_LIST_Table['VAX_LOT_LIST'].tolist()],
columns = index_columns, columns = index_columns,
index = VAX_LOT_LIST_Table.index) index = VAX_LOT_LIST_Table.index)

7
src/SymptomsByBatchcodesTableFactoryTest.py
View File

@@ -92,9 +92,10 @@ class SymptomsByBatchcodesTableFactoryTest(unittest.TestCase):
columns = ['SYMPTOMS'], columns = ['SYMPTOMS'],
data = [ ['Blood pressure orthostatic abnormal'], data = [ ['Blood pressure orthostatic abnormal'],
['Blood pressure orthostatic abnormal']], ['Blood pressure orthostatic abnormal']],
index = pd.MultiIndex.from_tuples( index = pd.Index(
names = ['VAX_LOT1', 'VAX_LOT2'], name = 'VAX_LOT1',
tuples = [['EW0175', str(np.nan)]] * 2)), data = ['EW0175',
'EW0175'])),
check_dtype = False) check_dtype = False)
def test_createSymptomsByBatchcodesTable_two_patients_distinct_symptoms(self): def test_createSymptomsByBatchcodesTable_two_patients_distinct_symptoms(self):