refactoring

This commit is contained in:
frankknoll
2022-03-02 12:07:12 +01:00
parent d73f84e4e8
commit 4e61397cd5

View File

@@ -22,23 +22,13 @@
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [],
"source": [ "source": [
"# download https://diviexchange.blob.core.windows.net/%24web/zeitreihe-tagesdaten.csv or https://www.intensivregister.de/#/aktuelle-lage/downloads\n", "def readTimeseries(download = False):\n",
"if False:\n", " timeSeriesFile = 'zeitreihe-tagesdaten.csv'\n",
" request.urlretrieve(\n", " if download:\n",
" 'https://diviexchange.blob.core.windows.net/%24web/zeitreihe-tagesdaten.csv',\n", " _downloadTimeseries(timeSeriesFile)\n",
" 'zeitreihe-tagesdaten.csv')\n" "\n",
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "f179762b",
"metadata": {},
"outputs": [],
"source": [
"def readTimeseries():\n",
" timeseries = pd.read_csv(\n", " timeseries = pd.read_csv(\n",
" 'zeitreihe-tagesdaten.csv',\n", " timeSeriesFile,\n",
" low_memory = False,\n", " low_memory = False,\n",
" usecols = ['date', 'bundesland', 'gemeindeschluessel', 'betten_belegt', 'betten_frei'],\n", " usecols = ['date', 'bundesland', 'gemeindeschluessel', 'betten_belegt', 'betten_frei'],\n",
" parse_dates = ['date'],\n", " parse_dates = ['date'],\n",
@@ -47,7 +37,24 @@
" 'gemeindeschluessel': 'string',\n", " 'gemeindeschluessel': 'string',\n",
" 'bundesland': 'string'\n", " 'bundesland': 'string'\n",
" })\n", " })\n",
" return timeseries.sort_values(by = 'date', ascending = True)\n" " return timeseries.sort_values(by = 'date', ascending = True)\n",
"\n",
"# download https://diviexchange.blob.core.windows.net/%24web/zeitreihe-tagesdaten.csv or https://www.intensivregister.de/#/aktuelle-lage/downloads\n",
"def _downloadTimeseries(timeSeriesFile):\n",
" request.urlretrieve(\n",
" 'https://diviexchange.blob.core.windows.net/%24web/zeitreihe-tagesdaten.csv',\n",
" timeSeriesFile)\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "3f992231",
"metadata": {},
"outputs": [],
"source": [
"timeSeries = readTimeseries(download = False)\n",
"timeSeries"
] ]
}, },
{ {
@@ -58,28 +65,34 @@
"outputs": [], "outputs": [],
"source": [ "source": [
"# download https://www.destatis.de/DE/Themen/Laender-Regionen/Regionales/Gemeindeverzeichnis/Administrativ/04-kreise.xlsx?__blob=publicationFile or https://www.destatis.de/DE/Themen/Laender-Regionen/Regionales/Gemeindeverzeichnis/Administrativ/04-kreise.html\n", "# download https://www.destatis.de/DE/Themen/Laender-Regionen/Regionales/Gemeindeverzeichnis/Administrativ/04-kreise.xlsx?__blob=publicationFile or https://www.destatis.de/DE/Themen/Laender-Regionen/Regionales/Gemeindeverzeichnis/Administrativ/04-kreise.html\n",
"if False:\n", "def readKreise(download = False):\n",
" request.urlretrieve(\n", " kreiseFile = '04-kreise.xlsx'\n",
" 'https://www.destatis.de/DE/Themen/Laender-Regionen/Regionales/Gemeindeverzeichnis/Administrativ/04-kreise.xlsx?__blob=publicationFile',\n", " if download:\n",
" '04-kreise.xlsx')" " _downloadKreise(kreiseFile)\n",
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "0927a6c2",
"metadata": {},
"outputs": [],
"source": [
"def readKreise():\n",
" kreise = pd.read_excel(\n", " kreise = pd.read_excel(\n",
" '04-kreise.xlsx',\n", " kreiseFile,\n",
" sheet_name = 'Kreisfreie Städte u. Landkreise',\n", " sheet_name = 'Kreisfreie Städte u. Landkreise',\n",
" header = 5,\n", " header = 5,\n",
" index_col = 0)\n", " index_col = 0)\n",
" kreise = kreise.rename(columns = {'2': 'Bundesland', 3: 'Kreis', 6: 'Einwohnerzahl'})[['Bundesland', 'Kreis', 'Einwohnerzahl']]\n", " kreise = kreise.rename(columns = {'2': 'Bundesland', 3: 'Kreis', 6: 'Einwohnerzahl'})[['Bundesland', 'Kreis', 'Einwohnerzahl']]\n",
" kreise.index.set_names(\"Key\", inplace = True)\n", " kreise.index.set_names(\"Key\", inplace = True)\n",
" return kreise" " return kreise\n",
"\n",
"def _downloadKreise(kreiseFile):\n",
" request.urlretrieve(\n",
" 'https://www.destatis.de/DE/Themen/Laender-Regionen/Regionales/Gemeindeverzeichnis/Administrativ/04-kreise.xlsx?__blob=publicationFile',\n",
" kreiseFile)\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "74ea4d55",
"metadata": {},
"outputs": [],
"source": [
"kreise = readKreise(download = False)\n",
"kreise"
] ]
}, },
{ {
@@ -119,7 +132,7 @@
"metadata": {}, "metadata": {},
"outputs": [], "outputs": [],
"source": [ "source": [
"timeSeries = ColumnsAdder(readKreise()).addKreisAndBundeslandAndEinwohnerzahlColumns(readTimeseries())\n", "timeSeries = ColumnsAdder(kreise).addKreisAndBundeslandAndEinwohnerzahlColumns(timeSeries)\n",
"timeSeries" "timeSeries"
] ]
}, },