refactoring

src/HowBadIsMyBatch.ipynb

@@ -24,6 +24,8 @@
     "from VAERSFileDownloader import updateVAERSFiles\n",
     "from GoogleAnalytics.CityCountsByBatchcodeTablesMerger import CityCountsByBatchcodeTablesMerger\n",
     "from IOUtils import IOUtils\n",
+    "from pathlib import Path\n",
+    "from GoogleAnalytics.Helper import persistCityCountsByClickedBatchcodeTables\n",
     "import os\n",
     "import pandas as pd\n",
     "\n",
@@ -268,30 +270,6 @@
     "cityCountsByClickedBatchcodeTable"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "6883ba3a",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "def dataframeWithoutIndexValue(dataframe, indexValue):\n",
-    "    return dataframe[dataframe.index != indexValue]\n",
-    "\n",
-    "def getMaxBatch(cityCountsByClickedBatchcodeTable):\n",
-    "    df_sorted = dataframeWithoutIndexValue(\n",
-    "        dataframe = (cityCountsByClickedBatchcodeTable\n",
-    "                        .groupby('VAX_LOT')\n",
-    "                        .sum()\n",
-    "                        .sort_values(by = 'CITY_COUNT_BY_VAX_LOT', ascending = False)),\n",
-    "        indexValue = '(not set)')\n",
-    "    print(df_sorted)\n",
-    "    return df_sorted.iloc[0].name\n",
-    "\n",
-    "maxBatch = getMaxBatch(cityCountsByClickedBatchcodeTable)\n",
-    "maxBatch"
-   ]
-  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -299,21 +277,10 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "cityCountsByClickedBatchcodeTable_maxBatch = cityCountsByClickedBatchcodeTable.loc[(maxBatch, slice(None), slice(None), slice(None)), :]\n",
-    "cityCountsByClickedBatchcodeTable_maxBatch"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "8d7a8bbf",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "(cityCountsByClickedBatchcodeTable_maxBatch\n",
-    " .reset_index()\n",
-    " .to_excel('tmp/cityCountsByClickedBatchcodeTable_maxBatch.xlsx'))\n",
-    "# import cityCountsByClickedBatchcodeTable_maxBatch.xlsx into https://www.google.com/mymaps"
+    "persistCityCountsByClickedBatchcodeTables(\n",
+    "    dataDir = Path('tmp/cityCountsByClickedBatchcodeTables'),\n",
+    "    n = 10,\n",
+    "    cityCountsByClickedBatchcodeTable = cityCountsByClickedBatchcodeTable)"
    ]
   },
   {