From fd3c49ce8b6290d852da82edbf994908a730a9ac Mon Sep 17 00:00:00 2001
From: frankknoll <Knoll_Frank@web.de>
Date: Tue, 25 Jan 2022 22:02:44 +0100
Subject: [PATCH] saveBatchCodeTable()

---
 .gitignore            |  1 +
 HowBadIsMyBatch.ipynb | 49 ++++++++++++++-----------------------------
 2 files changed, 17 insertions(+), 33 deletions(-)

diff --git a/.gitignore b/.gitignore
index 3b0fa91ee6e..474f82648ad 100644
--- a/.gitignore
+++ b/.gitignore
@@ -2,3 +2,4 @@ VAERS/
 .ipynb_checkpoints/
 VAERSDataUseGuide_en_September2021.pdf
 Vaccine Lot Numbers Expiration (1).xlsx
+results/
diff --git a/HowBadIsMyBatch.ipynb b/HowBadIsMyBatch.ipynb
index a68b001ff75..72b1b7dc362 100644
--- a/HowBadIsMyBatch.ipynb
+++ b/HowBadIsMyBatch.ipynb
@@ -44,17 +44,17 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "def createPivotTable(df):\n",
+    "def createBatchCodeTable(df):\n",
     "    def filter(df, col):\n",
     "        return df[df[col] == 'Y'][['VAX_LOT']]\n",
     "\n",
-    "    pivotTableDict = {\n",
+    "    batchCodeTableDict = {\n",
     "        'ADRs': df[['VAX_LOT']].value_counts(),\n",
     "        'DEATHS': filter(df, 'DIED').value_counts(),\n",
     "        'DISABILITIES': filter(df, 'DISABLE').value_counts(),\n",
     "        'LIFE THREATENING ILLNESSES': filter(df, 'L_THREAT').value_counts()\n",
     "    }\n",
-    "    return pd.concat(pivotTableDict, axis=1).replace(to_replace=np.nan, value=0)\n"
+    "    return pd.concat(batchCodeTableDict, axis=1).replace(to_replace=np.nan, value=0)\n"
    ]
   },
   {
@@ -64,7 +64,8 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "df_moderna = createDataFrame(\"VAERS\", \"MODERNA\")"
+    "def saveBatchCodeTable(manufacturer, excelFile):\n",
+    "    createBatchCodeTable(createDataFrame(\"VAERS\", manufacturer)).to_excel(excelFile)\n"
    ]
   },
   {
@@ -74,29 +75,9 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "df_moderna"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "d9191d12",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "pivotTable = createPivotTable(df_moderna)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "bb7b2963",
-   "metadata": {
-    "scrolled": false
-   },
-   "outputs": [],
-   "source": [
-    "pivotTable"
+    "saveBatchCodeTable(\"MODERNA\", \"results/moderna.xlsx\")\n",
+    "saveBatchCodeTable(\"PFIZER\\BIONTECH\", \"results/pfizer.xlsx\")\n",
+    "saveBatchCodeTable(\"JANSSEN\", \"results/janssen.xlsx\")"
    ]
   },
   {
@@ -121,12 +102,14 @@
     "\n",
     "class HowBadIsMyBatchTest(unittest.TestCase):\n",
     "\n",
-    "    def test_createPivotTable(self):\n",
+    "    def test_createBatchCodeTable(self):\n",
     "        # Given\n",
-    "        pivotTable = createPivotTable(createDataFrame(\"test/VAERS\", \"MODERNA\"))\n",
+    "        dataFrame = createDataFrame(\"test/VAERS\", \"MODERNA\")\n",
+    "        display(\"dataFrame:\", dataFrame)\n",
+    "        batchCodeTable = createBatchCodeTable(dataFrame)\n",
     "\n",
     "        # When\n",
-    "        pivotTableExpected = pd.DataFrame(\n",
+    "        batchCodeTableExpected = pd.DataFrame(\n",
     "            {\n",
     "                'ADRs': [2, 1],\n",
     "                'DEATHS': [0, 1],\n",
@@ -134,11 +117,11 @@
     "                'LIFE THREATENING ILLNESSES': [0.0, 0.0]\n",
     "            },\n",
     "            index = pd.MultiIndex.from_arrays([['025L20A', '037K20A']], names = ('VAX_LOT',)))\n",
-    "        display(\"actual:\", pivotTable)\n",
-    "        display(\"expected:\", pivotTableExpected)\n",
+    "        display(\"batchCodeTable:\", batchCodeTable)\n",
+    "        display(\"batchCodeTableExpected:\", batchCodeTableExpected)\n",
     "\n",
     "        # Then\n",
-    "        assert_frame_equal(pivotTable, pivotTableExpected, check_dtype = False)\n"
+    "        assert_frame_equal(batchCodeTable, batchCodeTableExpected, check_dtype = False)\n"
    ]
   },
   {