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

refactoring

Browse Source
This commit is contained in:
frankknoll
2022-02-06 10:56:33 +01:00
parent fa55ac1da4
commit 90d26d518d
1 changed files with 112 additions and 115 deletions
Download Patch File Download Diff File Expand all files Collapse all files

227
HowBadIsMyBatch.ipynb
View File

@@ -125,23 +125,20 @@
"\n",
"class DataFrameFilter:\n",
" \n",
" def __init__(self, dataFrame):\n",
" self.dataFrame = dataFrame\n",
" def filterByCovid19And(self, dataFrame, manufacturer = None, dose = None):\n",
" return dataFrame[self._isCovid19(dataFrame) & self._isManufacturer(dataFrame, manufacturer) & self._isDose(dataFrame, dose)]\n",
"\n",
" def filterByCovid19And(self, manufacturer = None, dose = None):\n",
" return self.dataFrame[self._isCovid19() & self._isManufacturer(manufacturer) & self._isDose(dose)]\n",
" def filterForSevereEffects(self, dataFrame, dose):\n",
" return self.filterByCovid19And(dataFrame, dose = dose)\n",
"\n",
" def filterForSevereEffects(self, dose):\n",
" return self.filterByCovid19And(dose = dose)\n",
" def _isCovid19(self, dataFrame):\n",
" return dataFrame[\"VAX_TYPE\"] == \"COVID19\"\n",
"\n",
" def _isCovid19(self):\n",
" return self.dataFrame[\"VAX_TYPE\"] == \"COVID19\"\n",
" def _isManufacturer(self, dataFrame, manufacturer):\n",
" return dataFrame[\"VAX_MANU\"] == manufacturer if manufacturer is not None else True\n",
"\n",
" def _isManufacturer(self, manufacturer):\n",
" return self.dataFrame[\"VAX_MANU\"] == manufacturer if manufacturer is not None else True\n",
"\n",
" def _isDose(self, dose):\n",
" return self.dataFrame[\"VAX_DOSE_SERIES\"].str.contains(dose) if dose is not None else True\n"
" def _isDose(self, dataFrame, dose):\n",
" return dataFrame[\"VAX_DOSE_SERIES\"].str.contains(dose) if dose is not None else True\n"
]
},
{
@@ -240,13 +237,13 @@
"\n",
" @staticmethod\n",
" def createBatchCodeTable(dataFrame : pd.DataFrame, manufacturer, dose):\n",
" filteredDataFrame = DataFrameFilter(dataFrame).filterByCovid19And(manufacturer = manufacturer, dose = dose)\n",
" filteredDataFrame = DataFrameFilter().filterByCovid19And(dataFrame, manufacturer = manufacturer, dose = dose)\n",
" return BatchCodeTableHelper(filteredDataFrame).createBatchCodeTable()\n",
"\n",
" # create table from https://www.howbadismybatch.com/combined.html\n",
" @staticmethod\n",
" def createSevereEffectsBatchCodeTable(dataFrame : pd.DataFrame, dose):\n",
" severeEffectsDataFrame = DataFrameFilter(dataFrame).filterForSevereEffects(dose)\n",
" severeEffectsDataFrame = DataFrameFilter().filterForSevereEffects(dataFrame, dose)\n",
" return BatchCodeTableHelper(severeEffectsDataFrame).createSevereEffectsBatchCodeTable()\n"
]
},
@@ -318,47 +315,47 @@
"\n",
" def test_filterByCovid19And(self):\n",
" # Given\n",
" dataFrameFilter = DataFrameFilter(\n",
" VaersDescr2DataFrameConverter.createDataFrameFromDescrs(\n",
" [\n",
" {\n",
" dataFrame = VaersDescr2DataFrameConverter.createDataFrameFromDescrs(\n",
" [\n",
" {\n",
" 'VAERSDATA': self.createDataFrame(\n",
" columns = ['DIED', 'L_THREAT', 'DISABLE'],\n",
" data = [ [1, 0, 0],\n",
" [0, 0, 1]],\n",
" index = [\n",
" \"0916600\",\n",
" \"0916601\"]),\n",
" 'VAERSVAX': self.createDataFrame(\n",
" columns = ['VAX_TYPE', 'VAX_MANU', 'VAX_LOT', 'VAX_DOSE_SERIES'],\n",
" data = [ ['COVID19', 'MODERNA', '037K20A', '1'],\n",
" ['COVID19', 'MODERNA', '025L20A', '1']],\n",
" index = [\n",
" \"0916600\",\n",
" \"0916601\"],\n",
" dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
" },\n",
" {\n",
" 'VAERSDATA': self.createDataFrame(\n",
" columns = ['DIED', 'L_THREAT', 'DISABLE'],\n",
" data = [ [1, 0, 0],\n",
" [0, 0, 1]],\n",
" columns = ['DIED', 'L_THREAT', 'DISABLE'],\n",
" data = [ [0, 0, 0],\n",
" [0, 0, 1]],\n",
" index = [\n",
" \"0916600\",\n",
" \"0916601\"]),\n",
" \"1996873\",\n",
" \"1996874\"]),\n",
" 'VAERSVAX': self.createDataFrame(\n",
" columns = ['VAX_TYPE', 'VAX_MANU', 'VAX_LOT', 'VAX_DOSE_SERIES'],\n",
" data = [ ['COVID19', 'MODERNA', '037K20A', '1'],\n",
" ['COVID19', 'MODERNA', '025L20A', '1']],\n",
" columns = ['VAX_TYPE', 'VAX_MANU', 'VAX_LOT', 'VAX_DOSE_SERIES'],\n",
" data = [ ['HPV9', 'MERCK & CO. INC.', 'R017624', 'UNK'],\n",
" ['COVID19', 'MODERNA', '025L20A', '1']],\n",
" index = [\n",
" \"0916600\",\n",
" \"0916601\"],\n",
" dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
" },\n",
" {\n",
" 'VAERSDATA': self.createDataFrame(\n",
" columns = ['DIED', 'L_THREAT', 'DISABLE'],\n",
" data = [ [0, 0, 0],\n",
" [0, 0, 1]],\n",
" index = [\n",
" \"1996873\",\n",
" \"1996874\"]),\n",
" 'VAERSVAX': self.createDataFrame(\n",
" columns = ['VAX_TYPE', 'VAX_MANU', 'VAX_LOT', 'VAX_DOSE_SERIES'],\n",
" data = [ ['HPV9', 'MERCK & CO. INC.', 'R017624', 'UNK'],\n",
" ['COVID19', 'MODERNA', '025L20A', '1']],\n",
" index = [\n",
" \"1996873\",\n",
" \"1996874\"],\n",
" dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
" }\n",
" ]))\n",
" \"1996874\"],\n",
" dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
" }\n",
" ])\n",
" dataFrameFilter = DataFrameFilter()\n",
" \n",
" # When\n",
" dataFrame = dataFrameFilter.filterByCovid19And(manufacturer = \"MODERNA\", dose = '1')\n",
" dataFrame = dataFrameFilter.filterByCovid19And(dataFrame, manufacturer = \"MODERNA\", dose = '1')\n",
" \n",
" # Then\n",
" dataFrameExpected = self.createDataFrame(\n",
@@ -375,30 +372,30 @@
"\n",
" def test_filterForSevereEffects(self):\n",
" # Given\n",
" dataFrameFilter = DataFrameFilter(\n",
" VaersDescr2DataFrameConverter.createDataFrameFromDescrs(\n",
" [\n",
" {\n",
" 'VAERSDATA': self.createDataFrame(\n",
" columns = ['DIED', 'L_THREAT', 'DISABLE', 'HOSPITAL', 'ER_VISIT'],\n",
" data = [ [1, 1, 0, 1, 1],\n",
" [0, 0, 1, 0, 1]],\n",
" index = [\n",
" \"0916600\",\n",
" \"0916601\"]),\n",
" 'VAERSVAX': self.createDataFrame(\n",
" columns = ['VAX_TYPE', 'VAX_MANU', 'VAX_LOT', 'VAX_DOSE_SERIES'],\n",
" data = [ ['COVID19', 'MODERNA', '037K20A', '1'],\n",
" ['COVID19', 'PFIZER\\BIONTECH', '025L20A', '1']],\n",
" index = [\n",
" \"0916600\",\n",
" \"0916601\"],\n",
" dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
" }\n",
" ]))\n",
" dataFrame = VaersDescr2DataFrameConverter.createDataFrameFromDescrs(\n",
" [\n",
" {\n",
" 'VAERSDATA': self.createDataFrame(\n",
" columns = ['DIED', 'L_THREAT', 'DISABLE', 'HOSPITAL', 'ER_VISIT'],\n",
" data = [ [1, 1, 0, 1, 1],\n",
" [0, 0, 1, 0, 1]],\n",
" index = [\n",
" \"0916600\",\n",
" \"0916601\"]),\n",
" 'VAERSVAX': self.createDataFrame(\n",
" columns = ['VAX_TYPE', 'VAX_MANU', 'VAX_LOT', 'VAX_DOSE_SERIES'],\n",
" data = [ ['COVID19', 'MODERNA', '037K20A', '1'],\n",
" ['COVID19', 'PFIZER\\BIONTECH', '025L20A', '1']],\n",
" index = [\n",
" \"0916600\",\n",
" \"0916601\"],\n",
" dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
" }\n",
" ])\n",
" dataFrameFilter = DataFrameFilter()\n",
"\n",
" # When\n",
" dataFrame = dataFrameFilter.filterForSevereEffects(dose = '1')\n",
" dataFrame = dataFrameFilter.filterForSevereEffects(dataFrame, dose = '1')\n",
" \n",
" # Then\n",
" dataFrameExpected = self.createDataFrame(\n",
@@ -413,28 +410,28 @@
"\n",
" def test_filterByFirstDose(self):\n",
" # Given\n",
" dataFrameFilter = DataFrameFilter(\n",
" VaersDescr2DataFrameConverter.createDataFrameFromDescrs(\n",
" [\n",
" {\n",
" 'VAERSDATA': self.createDataFrame(\n",
" columns = ['DIED', 'L_THREAT', 'DISABLE'],\n",
" data = [ [1, 0, 0]],\n",
" index = [\n",
" \"1048786\"]),\n",
" 'VAERSVAX': self.createDataFrame(\n",
" columns = ['VAX_TYPE', 'VAX_MANU', 'VAX_LOT', 'VAX_DOSE_SERIES'],\n",
" data = [ ['COVID19', 'MODERNA', '016M20A', '2'],\n",
" ['COVID19', 'MODERNA', '030L20A', '1']],\n",
" index = [\n",
" \"1048786\",\n",
" \"1048786\"],\n",
" dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
" }\n",
" ]))\n",
" dataFrame = VaersDescr2DataFrameConverter.createDataFrameFromDescrs(\n",
" [\n",
" {\n",
" 'VAERSDATA': self.createDataFrame(\n",
" columns = ['DIED', 'L_THREAT', 'DISABLE'],\n",
" data = [ [1, 0, 0]],\n",
" index = [\n",
" \"1048786\"]),\n",
" 'VAERSVAX': self.createDataFrame(\n",
" columns = ['VAX_TYPE', 'VAX_MANU', 'VAX_LOT', 'VAX_DOSE_SERIES'],\n",
" data = [ ['COVID19', 'MODERNA', '016M20A', '2'],\n",
" ['COVID19', 'MODERNA', '030L20A', '1']],\n",
" index = [\n",
" \"1048786\",\n",
" \"1048786\"],\n",
" dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
" }\n",
" ])\n",
" dataFrameFilter = DataFrameFilter()\n",
" \n",
" # When\n",
" dataFrame = dataFrameFilter.filterByCovid19And(manufacturer = \"MODERNA\", dose = '1')\n",
" dataFrame = dataFrameFilter.filterByCovid19And(dataFrame, manufacturer = \"MODERNA\", dose = '1')\n",
" \n",
" # Then\n",
" dataFrameExpected = self.createDataFrame(\n",
@@ -447,28 +444,28 @@
"\n",
" def test_filterBySecondDose(self):\n",
" # Given\n",
" dataFrameFilter = DataFrameFilter(\n",
" VaersDescr2DataFrameConverter.createDataFrameFromDescrs(\n",
" [\n",
" {\n",
" 'VAERSDATA': self.createDataFrame(\n",
" columns = ['DIED', 'L_THREAT', 'DISABLE'],\n",
" data = [ [1, 0, 0]],\n",
" index = [\n",
" \"1048786\"]),\n",
" 'VAERSVAX': self.createDataFrame(\n",
" columns = ['VAX_TYPE', 'VAX_MANU', 'VAX_LOT', 'VAX_DOSE_SERIES'],\n",
" data = [ ['COVID19', 'MODERNA', '016M20A', '2'],\n",
" ['COVID19', 'MODERNA', '030L20A', '1']],\n",
" index = [\n",
" \"1048786\",\n",
" \"1048786\"],\n",
" dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
" }\n",
" ]))\n",
" dataFrame = VaersDescr2DataFrameConverter.createDataFrameFromDescrs(\n",
" [\n",
" {\n",
" 'VAERSDATA': self.createDataFrame(\n",
" columns = ['DIED', 'L_THREAT', 'DISABLE'],\n",
" data = [ [1, 0, 0]],\n",
" index = [\n",
" \"1048786\"]),\n",
" 'VAERSVAX': self.createDataFrame(\n",
" columns = ['VAX_TYPE', 'VAX_MANU', 'VAX_LOT', 'VAX_DOSE_SERIES'],\n",
" data = [ ['COVID19', 'MODERNA', '016M20A', '2'],\n",
" ['COVID19', 'MODERNA', '030L20A', '1']],\n",
" index = [\n",
" \"1048786\",\n",
" \"1048786\"],\n",
" dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
" }\n",
" ])\n",
" dataFrameFilter = DataFrameFilter()\n",
"\n",
" # When\n",
" dataFrame = dataFrameFilter.filterByCovid19And(manufacturer = \"MODERNA\", dose = '2')\n",
" dataFrame = dataFrameFilter.filterByCovid19And(dataFrame, manufacturer = \"MODERNA\", dose = '2')\n",
" \n",
" # Then\n",
" dataFrameExpected = self.createDataFrame(\n",
@@ -621,7 +618,7 @@
" \"1048786\",\n",
" \"4711\"],\n",
" dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
" dataFrame = DataFrameFilter(dataFrame).filterByCovid19And()\n",
" dataFrame = DataFrameFilter().filterByCovid19And(dataFrame)\n",
" \n",
" # When\n",
" doseTable = DoseAnalysis.getDoseTable(dataFrame)\n",
@@ -652,7 +649,7 @@
" \"1048786\",\n",
" \"4711\"],\n",
" dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
" dataFrame = DataFrameFilter(dataFrame).filterByCovid19And()\n",
" dataFrame = DataFrameFilter().filterByCovid19And(dataFrame)\n",
" \n",
" # When\n",
" doseByMonthTable = DoseAnalysis.getDoseByMonthTable(dataFrame)\n",
@@ -815,13 +812,13 @@
"def getDoseTable():\n",
" vaersDescrs = VaersDescrReader(dataDir = \"VAERS\").readAllVaersDescrs()\n",
" dataFrame = VaersDescr2DataFrameConverter.createDataFrameFromDescrs(vaersDescrs)\n",
" dataFrame = DataFrameFilter(dataFrame).filterByCovid19And()\n",
" dataFrame = DataFrameFilter().filterByCovid19And(dataFrame)\n",
" return DoseAnalysis.getDoseTable(dataFrame)\n",
"\n",
"def getDoseByMonthTable():\n",
" vaersDescrs = VaersDescrReader(dataDir = \"VAERS\").readAllVaersDescrs()\n",
" dataFrame = VaersDescr2DataFrameConverter.createDataFrameFromDescrs(vaersDescrs)\n",
" dataFrame = DataFrameFilter(dataFrame).filterByCovid19And()\n",
" dataFrame = DataFrameFilter().filterByCovid19And(dataFrame)\n",
" return DoseAnalysis.getDoseByMonthTable(dataFrame)"
]
},