refactoring

This commit is contained in:
frankknoll
2022-01-28 17:50:19 +01:00
parent 3a4d44f8d0
commit 08ad895d3a

View File

@@ -36,19 +36,19 @@
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [],
"source": [ "source": [
"def read_csv(file, usecols):\n",
" return pd.read_csv(file, index_col = 'VAERS_ID', encoding = 'latin1', low_memory = False, usecols = usecols)\n",
"\n",
"def readVaersDescr(dataDir, year):\n",
" folder = dataDir + \"/\" + year + \"VAERSData/\"\n",
" return {\n",
" 'VAERSDATA': read_csv(folder + year + \"VAERSDATA.csv\", ['VAERS_ID', 'DIED', 'L_THREAT', 'DISABLE']),\n",
" 'VAERSVAX': read_csv(folder + year + \"VAERSVAX.csv\", ['VAERS_ID', 'VAX_TYPE', 'VAX_MANU', 'VAX_LOT'])\n",
" }\n",
"\n",
"def createDataFrame(dataDir, manufacturer):\n", "def createDataFrame(dataDir, manufacturer):\n",
" def readVaersDescr(year):\n",
" def read_csv(file, usecols):\n",
" return pd.read_csv(file, index_col = 'VAERS_ID', encoding = 'latin1', low_memory = False, usecols = usecols)\n",
"\n",
" folder = dataDir + \"/\" + year + \"VAERSData/\"\n",
" return {\n",
" 'VAERSDATA': read_csv(folder + year + \"VAERSDATA.csv\", ['VAERS_ID', 'DIED', 'L_THREAT', 'DISABLE']),\n",
" 'VAERSVAX': read_csv(folder + year + \"VAERSVAX.csv\", ['VAERS_ID', 'VAX_TYPE', 'VAX_MANU', 'VAX_LOT'])\n",
" }\n",
"\n",
" return _createDataFrame(\n", " return _createDataFrame(\n",
" [readVaersDescr(dataDir, \"2021\"), readVaersDescr(dataDir, \"2022\")],\n", " [readVaersDescr(\"2021\"), readVaersDescr(\"2022\")],\n",
" manufacturer)" " manufacturer)"
] ]
}, },