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
2022-01-29 15:03:37 +01:00
parent ae7cb41dac
commit d860b2d630
1 changed files with 43 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
HowBadIsMyBatch.ipynb
View File

@@ -21,16 +21,16 @@
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [],
"source": [ "source": [
"def createDataFrameFromDescr(vaersDescr, manufacturer, dose):\n", "def filter(df, manufacturer, dose):\n",
" def filter(df):\n",
" return df[\n", " return df[\n",
" (df[\"VAX_TYPE\"] == \"COVID19\") &\n", " (df[\"VAX_TYPE\"] == \"COVID19\") &\n",
" (df[\"VAX_MANU\"] == manufacturer) &\n", " (df[\"VAX_MANU\"] == manufacturer) &\n",
" (df[\"VAX_DOSE_SERIES\"].str.contains(dose))]\n", " (df[\"VAX_DOSE_SERIES\"].str.contains(dose))]\n",
"\n", "\n",
"def createDataFrameFromDescr(vaersDescr, manufacturer, dose):\n",
" return pd.merge(\n", " return pd.merge(\n",
" vaersDescr['VAERSDATA'],\n", " vaersDescr['VAERSDATA'],\n",
" filter(vaersDescr['VAERSVAX']),\n", " filter(vaersDescr['VAERSVAX'], manufacturer, dose),\n",
" left_index = True,\n", " left_index = True,\n",
" right_index = True)\n", " right_index = True)\n",
"\n", "\n",
@@ -47,8 +47,6 @@
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [],
"source": [ "source": [
"def createDataFrameFromFiles(dataDir, manufacturer, dose):\n",
" def readVaersDescr(year):\n",
"def read_csv(file, usecols, dtype = {}):\n", "def read_csv(file, usecols, dtype = {}):\n",
" return pd.read_csv(\n", " return pd.read_csv(\n",
" file,\n", " file,\n",
@@ -58,6 +56,7 @@
" usecols = usecols,\n", " usecols = usecols,\n",
" dtype = dtype)\n", " dtype = dtype)\n",
"\n", "\n",
"def readVaersDescr(dataDir, year):\n",
" folder = dataDir + \"/\" + year + \"VAERSData/\"\n", " folder = dataDir + \"/\" + year + \"VAERSData/\"\n",
" return {\n", " return {\n",
" 'VAERSDATA':\n", " 'VAERSDATA':\n",
@@ -71,8 +70,9 @@
" dtype = {\"VAX_DOSE_SERIES\": \"string\"})\n", " dtype = {\"VAX_DOSE_SERIES\": \"string\"})\n",
" }\n", " }\n",
"\n", "\n",
"def createDataFrameFromFiles(dataDir, manufacturer, dose):\n",
" return createDataFrameFromDescrs(\n", " return createDataFrameFromDescrs(\n",
" [readVaersDescr(\"2021\"), readVaersDescr(\"2022\")],\n", " [readVaersDescr(dataDir, \"2021\"), readVaersDescr(dataDir, \"2022\")],\n",
" manufacturer,\n", " manufacturer,\n",
" dose)" " dose)"
] ]
@@ -110,6 +110,7 @@
{ {
"cell_type": "code", "cell_type": "code",
"execution_count": null, "execution_count": null,
"id": "e59a1825",
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [],
"source": [ "source": [