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

removeUnknownBatchCodes

Browse Source
This commit is contained in:
frankknoll
2022-02-22 22:59:23 +01:00
parent 43403e7872
commit 89c19d4d6d
82 changed files with 120 additions and 82 deletions
Download Patch File Download Diff File Expand all files Collapse all files

40
src/HowBadIsMyBatch.ipynb
View File

@@ -103,6 +103,7 @@
" \n",
" @staticmethod\n",
" def normalize(dataFrame):\n",
" DataFrameNormalizer.removeUnknownBatchCodes(dataFrame)\n",
" DataFrameNormalizer.convertVAX_LOTColumnToUpperCase(dataFrame)\n",
" DataFrameNormalizer._convertColumnsOfDataFrame_Y_to_1_else_0(\n",
" dataFrame,\n",
@@ -113,6 +114,14 @@
" dataFrame['VAX_LOT'] = dataFrame['VAX_LOT'].str.upper()\n",
"\n",
" @staticmethod\n",
" def removeUnknownBatchCodes(dataFrame):\n",
" dataFrame.drop(DataFrameNormalizer._isUnknownBatchCode(dataFrame).index, inplace = True)\n",
"\n",
" @staticmethod\n",
" def _isUnknownBatchCode(dataFrame):\n",
" return dataFrame[dataFrame['VAX_LOT'].str.contains(pat = 'UNKNOWN', regex = False, case = False, na = False)]\n",
"\n",
" @staticmethod\n",
" def _convertColumnsOfDataFrame_Y_to_1_else_0(dataFrame, columns):\n",
" for column in columns:\n",
" DataFrameNormalizer._convertColumnOfDataFrame_Y_to_1_else_0(dataFrame, column)\n",
@@ -422,7 +431,36 @@
" \"0916600\",\n",
" \"0916601\",\n",
" \"1996874\"])\n",
" assert_frame_equal(dataFrame, dataFrameExpected, check_dtype = False)\n"
" assert_frame_equal(dataFrame, dataFrameExpected, check_dtype = False)\n",
"\n",
" def test_removeUnknownBatchCodes(self):\n",
" # Given\n",
" dataFrame = TestHelper.createDataFrame(\n",
" columns = ['VAX_LOT'],\n",
" data = [ ['UNKNOWN'],\n",
" ['N/A Unknown'],\n",
" [np.nan],\n",
" ['UNKNOWN TO ME'],\n",
" ['030L20B']],\n",
" index = [\n",
" \"1048786\",\n",
" \"1048786\",\n",
" \"123\",\n",
" \"4711\",\n",
" \"0815\"])\n",
" \n",
" # When\n",
" DataFrameNormalizer.removeUnknownBatchCodes(dataFrame)\n",
" \n",
" # Then\n",
" dataFrameExpected = TestHelper.createDataFrame(\n",
" columns = ['VAX_LOT'],\n",
" data = [ [np.nan],\n",
" ['030L20B']],\n",
" index = [\n",
" \"123\",\n",
" \"0815\"])\n",
" assert_frame_equal(dataFrame, dataFrameExpected, check_dtype = False)"
]
},
{