refactoring

.gitignore

@@ -1,5 +1,4 @@
 VAERS/
 .ipynb_checkpoints/
-VAERSDataUseGuide_en_September2021.pdf
+config/
-Vaccine Lot Numbers Expiration (1).xlsx
 results/

HowBadIsMyBatch.ipynb

@@ -106,21 +106,21 @@
     "        # Given\n",
     "        dataFrame = createDataFrame(\"test/VAERS\", \"MODERNA\")\n",
     "        display(\"dataFrame:\", dataFrame)\n",
-    "        batchCodeTable = createBatchCodeTable(dataFrame)\n",
     "\n",
     "        # When\n",
+    "        batchCodeTable = createBatchCodeTable(dataFrame)\n",
+    "        display(\"batchCodeTable:\", batchCodeTable)\n",
+    "\n",
+    "        # Then\n",
     "        batchCodeTableExpected = pd.DataFrame(\n",
     "            {\n",
     "                'ADRs': [2, 1],\n",
     "                'DEATHS': [0, 1],\n",
     "                'DISABILITIES': [2, 0],\n",
-    "                'LIFE THREATENING ILLNESSES': [0.0, 0.0]\n",
+    "                'LIFE THREATENING ILLNESSES': [0, 0]\n",
     "            },\n",
     "            index = pd.MultiIndex.from_arrays([['025L20A', '037K20A']], names = ('VAX_LOT',)))\n",
-    "        display(\"batchCodeTable:\", batchCodeTable)\n",
     "        display(\"batchCodeTableExpected:\", batchCodeTableExpected)\n",
-    "\n",
-    "        # Then\n",
     "        assert_frame_equal(batchCodeTable, batchCodeTableExpected, check_dtype = False)\n"
    ]
   },

README.md

@@ -1,30 +1,2 @@
 # HowBadIsMyBatch
 
-jupyter notebook
-
-FK-TODO:
-- VAX_LOT-Spalte normalisieren, d.h. toUpperCase(), Format des jeweiligen Herstellers berücksichtigen und "verschmutzte" Einträge säubern, denn sie stellen alle dieselbe Charge dar:
-039k20a
-MOD039K20A
-#039K20A
-039K20A-MODERNA
-039K20A-2A
-039K20A or 039L
-Moderna/039K20A
-MODERNA 039K20A
-MODERNA039K20A
-Modena 039k20A
-L039K20A
-M039K20A
-MOD; 039K20A
-m0039k20A
-u039k20a
-6/21 039K20A
-2039K20A
-013L20A 039K20A#039K20A
-#039K
-039K20A  12-31-
-039K20A & 031M2
-039K20A and 032
-039K20A, 011L20
-

help.txt

@@ -0,0 +1,27 @@
+jupyter notebook
+
+FK-TODO:
+- VAX_LOT-Spalte normalisieren, d.h. toUpperCase(), Format des jeweiligen Herstellers berücksichtigen und "verschmutzte" Einträge säubern, denn sie stellen alle dieselbe Charge dar:
+039k20a
+MOD039K20A
+#039K20A
+039K20A-MODERNA
+039K20A-2A
+039K20A or 039L
+Moderna/039K20A
+MODERNA 039K20A
+MODERNA039K20A
+Modena 039k20A
+L039K20A
+M039K20A
+MOD; 039K20A
+m0039k20A
+u039k20a
+6/21 039K20A
+2039K20A
+013L20A 039K20A#039K20A
+#039K
+039K20A  12-31-
+039K20A & 031M2
+039K20A and 032
+039K20A, 011L20