From 38170520a1754c5f3d0da522182c922fd5e32555 Mon Sep 17 00:00:00 2001
From: frankknoll <Knoll_Frank@web.de>
Date: Sun, 30 Jan 2022 23:34:56 +0100
Subject: [PATCH] refactoring

---
 HowBadIsMyBatch.ipynb | 60 +++++++++++++++++++++++++++++++++++++------
 1 file changed, 52 insertions(+), 8 deletions(-)

diff --git a/HowBadIsMyBatch.ipynb b/HowBadIsMyBatch.ipynb
index 673fe4d95e0..1921ab0a25e 100644
--- a/HowBadIsMyBatch.ipynb
+++ b/HowBadIsMyBatch.ipynb
@@ -21,12 +21,6 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "def filterDataFrame(df, manufacturer, dose):\n",
-    "    return df[\n",
-    "        (df[\"VAX_TYPE\"] == \"COVID19\") &\n",
-    "        (df[\"VAX_MANU\"] == manufacturer) &\n",
-    "        (df[\"VAX_DOSE_SERIES\"].str.contains(dose))]\n",
-    "\n",
     "def createDataFrameFromDescr(vaersDescr):\n",
     "    return pd.merge(\n",
     "            vaersDescr['VAERSDATA'],\n",
@@ -41,7 +35,21 @@
     "    return pd.concat(dataFrames)\n",
     "\n",
     "def createAndFilterDataFrameFromDescrs(vaersDescrs, manufacturer, dose):\n",
-    "    return filterDataFrame(createDataFrameFromDescrs(vaersDescrs), manufacturer, dose)\n"
+    "    def filterDataFrame(df):\n",
+    "        return df[\n",
+    "            (df[\"VAX_TYPE\"] == \"COVID19\") &\n",
+    "            (df[\"VAX_MANU\"] == manufacturer) &\n",
+    "            (df[\"VAX_DOSE_SERIES\"].str.contains(dose))]\n",
+    "    \n",
+    "    return filterDataFrame(createDataFrameFromDescrs(vaersDescrs))\n",
+    "\n",
+    "def createDataFrameSevereEffectsFromDescrs(vaersDescrs, dose):\n",
+    "    def filterDataFrame(df):\n",
+    "        return df[\n",
+    "            (df[\"VAX_TYPE\"] == \"COVID19\") &\n",
+    "            (df[\"VAX_DOSE_SERIES\"].str.contains(dose))]\n",
+    "\n",
+    "    return filterDataFrame(createDataFrameFromDescrs(vaersDescrs))"
    ]
   },
   {
@@ -66,7 +74,7 @@
     "        'VAERSDATA':\n",
     "            read_csv(\n",
     "                folder + year + \"VAERSDATA.csv\",\n",
-    "                ['VAERS_ID', 'DIED', 'L_THREAT', 'DISABLE']),\n",
+    "                ['VAERS_ID', 'DIED', 'L_THREAT', 'DISABLE', 'HOSPITAL', 'ER_VISIT']),\n",
     "        'VAERSVAX':\n",
     "                read_csv(\n",
     "                    folder + year + \"VAERSVAX.csv\",\n",
@@ -177,6 +185,42 @@
     "            dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
     "        assert_frame_equal(dataFrame, dataFrameExpected, check_dtype = False)\n",
     "\n",
+    "    def test_createDataFrameFromForSevereEffects(self):\n",
+    "        # Given\n",
+    "        vaersDescrs = [\n",
+    "               {\n",
+    "                   'VAERSDATA': self.createDataFrame(\n",
+    "                       columns = ['DIED', 'L_THREAT', 'DISABLE', 'HOSPITAL', 'ER_VISIT'],\n",
+    "                       data = [  ['Y',    'Y',        np.NaN,    'Y',        'Y'],\n",
+    "                                 [np.NaN, np.NaN,     'Y',       np.NaN,     'Y']],\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",
+    "            \n",
+    "        # When\n",
+    "        dataFrame = createDataFrameSevereEffectsFromDescrs(vaersDescrs, '1')\n",
+    "        \n",
+    "        # Then\n",
+    "        dataFrameExpected = self.createDataFrame(\n",
+    "            columns = ['DIED', 'L_THREAT', 'DISABLE', 'HOSPITAL', 'ER_VISIT', 'VAX_TYPE', 'VAX_MANU',        'VAX_LOT', 'VAX_DOSE_SERIES'],\n",
+    "            data = [  ['Y',    'Y',        np.NaN,    'Y',        'Y',        'COVID19',  'MODERNA',         '037K20A', '1'],\n",
+    "                      [np.NaN,  np.NaN,    'Y',       np.NaN,     'Y',        'COVID19',  'PFIZER\\BIONTECH', '025L20A', '1']],\n",
+    "            index = [\n",
+    "                \"0916600\",\n",
+    "                \"0916601\"],\n",
+    "            dtypes = {'VAX_DOSE_SERIES': \"string\"})\n",
+    "        assert_frame_equal(dataFrame, dataFrameExpected, check_dtype = False)\n",
+    "\n",
     "    def test_createAndFilterDataFrameFromDescrsWithFirstDose(self):\n",
     "        # Given\n",
     "        vaersDescrs = [\n",