adding CityCountsByBatchcodeTablesMerger

src/GoogleAnalytics/CityCountsByBatchcodeTablesMerger.py

@@ -0,0 +1,19 @@
+import pandas as pd
+from GoogleAnalytics.RegionCountsByClickedBatchcodeProvider import RegionCountsByClickedBatchcodeProvider
+from GoogleAnalytics.FilesProvider import FilesProvider
+from GoogleAnalytics.Resolution import Resolution
+
+class CityCountsByBatchcodeTablesMerger:
+
+    @staticmethod
+    def getCityCountsByClickedBatchcode(dataDir):
+        files = FilesProvider(dataDir).getFilesHavingResolution(Resolution.CITY)
+        cityCountsByClickedBatchcodeTables = [RegionCountsByClickedBatchcodeProvider._getCityCountsByClickedBatchcode(file) for file in files]
+        table = pd.concat(cityCountsByClickedBatchcodeTables)
+        return CityCountsByBatchcodeTablesMerger._getCityCountsByClickedBatchcodeFromTable(table)
+
+    @staticmethod
+    def _getCityCountsByClickedBatchcodeFromTable(cityCountsByClickedBatchcodeTable):
+        return (cityCountsByClickedBatchcodeTable
+                .groupby(cityCountsByClickedBatchcodeTable.index.names)
+                .sum())

src/GoogleAnalytics/CityCountsByBatchcodeTablesMergerTest.py

@@ -0,0 +1,27 @@
+import unittest
+from pandas.testing import assert_frame_equal
+from TestHelper import TestHelper
+import pandas as pd
+from GoogleAnalytics.CityCountsByBatchcodeTablesMerger import CityCountsByBatchcodeTablesMerger
+
+class CityCountsByBatchcodeTablesMergerTest(unittest.TestCase):
+
+    def test_getCityCountsByClickedBatchcode(self):
+        # Given
+
+        # When
+        cityCountsByClickedBatchcodeTable = CityCountsByBatchcodeTablesMerger.getCityCountsByClickedBatchcode('src/testdata/GoogleAnalytics')
+        
+        # Then
+        assert_frame_equal(
+            cityCountsByClickedBatchcodeTable,
+            TestHelper.createDataFrame(
+                columns = ['CITY_COUNT_BY_VAX_LOT'],
+                data = [  [100 + 200],
+                          [10 + 20],
+                          [20 + 40]],
+                index = pd.MultiIndex.from_tuples(
+                            names =   ['VAX_LOT',  'COUNTRY',       'REGION',     'CITY'],
+                            tuples = [['#003B21A', 'United States', 'California', 'Roseville'],
+                                      ['000086A',  'Germany',       'Bavaria',    'Nordlingen'],
+                                      ['000086A',  'Germany',       'Bavaria',    'Nuremberg']])))

src/GoogleAnalytics/RegionCountsByBatchcodeTablesMerger.py

@@ -1,13 +1,13 @@
+import pandas as pd
 from GoogleAnalytics.RegionCountsByClickedBatchcodeProvider import RegionCountsByClickedBatchcodeProvider
 from GoogleAnalytics.FilesProvider import FilesProvider
 from GoogleAnalytics.Resolution import Resolution
-from TablesHelper import TablesHelper
 
 class RegionCountsByBatchcodeTablesMerger:
 
     @staticmethod
     def getRegionCountsByClickedBatchcode(dataDir):
         files = FilesProvider(dataDir).getFilesHavingResolution(Resolution.CITY)
-        tables = [RegionCountsByClickedBatchcodeProvider.getRegionCountsByClickedBatchcode(file) for file in files]
-        table = TablesHelper.concatTables_groupByIndex_sum(tables)
-        return table
+        cityCountsByClickedBatchcodeTables = [RegionCountsByClickedBatchcodeProvider._getCityCountsByClickedBatchcode(file) for file in files]
+        table = pd.concat(cityCountsByClickedBatchcodeTables)
+        return RegionCountsByClickedBatchcodeProvider._getRegionCountsByClickedBatchcodeFromTable(table)

src/GoogleAnalytics/RegionCountsByClickedBatchcodeProvider.py

@@ -5,13 +5,7 @@ class RegionCountsByClickedBatchcodeProvider:
 
     @staticmethod
     def getRegionCountsByClickedBatchcode(file):
-        cityCountsByClickedBatchcodeTable = RegionCountsByClickedBatchcodeProvider._getCityCountsByClickedBatchcode(file)
-        return (cityCountsByClickedBatchcodeTable
-                .groupby(['VAX_LOT', 'COUNTRY', 'REGION'])
-                .agg(REGION_COUNT_BY_VAX_LOT =
-                      pd.NamedAgg(
-                          column = 'CITY_COUNT_BY_VAX_LOT',
-                          aggfunc = sum)))
+        return RegionCountsByClickedBatchcodeProvider._getRegionCountsByClickedBatchcodeFromTable(RegionCountsByClickedBatchcodeProvider._getCityCountsByClickedBatchcode(file))
 
     # FK-TODO: delegate same method CountryCountsByClickedBatchcodeProvider._getCityCountsByClickedBatchcode() to here
     @staticmethod
@@ -25,3 +19,12 @@ class RegionCountsByClickedBatchcodeProvider:
                 'Event count': 'CITY_COUNT_BY_VAX_LOT'
             },
             index_columns = ['COUNTRY', 'REGION', 'CITY'])
+
+    @staticmethod
+    def _getRegionCountsByClickedBatchcodeFromTable(cityCountsByClickedBatchcodeTable):
+        return (cityCountsByClickedBatchcodeTable
+                .groupby(['VAX_LOT', 'COUNTRY', 'REGION'])
+                .agg(REGION_COUNT_BY_VAX_LOT =
+                      pd.NamedAgg(
+                          column = 'CITY_COUNT_BY_VAX_LOT',
+                          aggfunc = sum)))

src/HowBadIsMyBatch.ipynb

@@ -249,6 +249,14 @@
     "    '../docs/data/barChartDescriptionTable.json')\n"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "c5a66a3b",
+   "metadata": {},
+   "source": [
+    "# Google Analytics"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -283,6 +291,78 @@
     "regionCountsByClickedBatchcodeTable4Germany.to_excel('tmp/regionCountsByClickedBatchcodeTable4Germany.xlsx')"
    ]
   },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "3c05fcfc",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# VAX_LOT: EX8679\n",
+    "(regionCountsByClickedBatchcodeTable4Germany\n",
+    "    .groupby('VAX_LOT')\n",
+    "    .sum()\n",
+    "    .sort_values(by = 'REGION_COUNT_BY_VAX_LOT', ascending = False))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "370bf329",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "(regionCountsByClickedBatchcodeTable4Germany\n",
+    "    .loc[('EX8679', slice(None), slice(None)), :]\n",
+    "    .sort_values(by = 'REGION_COUNT_BY_VAX_LOT', ascending = False))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "9740c40b",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from GoogleAnalytics.CityCountsByBatchcodeTablesMerger import CityCountsByBatchcodeTablesMerger\n",
+    "\n",
+    "cityCountsByClickedBatchcodeTable = CityCountsByBatchcodeTablesMerger.getCityCountsByClickedBatchcode('data/GoogleAnalytics')\n",
+    "cityCountsByClickedBatchcodeTable"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "0dac0ea6",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "cityCountsByClickedBatchcodeTable_EX8679_Germany = cityCountsByClickedBatchcodeTable.loc[('EX8679', 'Germany', slice(None), slice(None)), :]\n",
+    "cityCountsByClickedBatchcodeTable_EX8679_Germany"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "8d7a8bbf",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "cityCountsByClickedBatchcodeTable_EX8679_Germany.to_excel('tmp/cityCountsByClickedBatchcodeTable_EX8679_Germany.xlsx')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "494943f3",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "(cityCountsByClickedBatchcodeTable_EX8679_Germany\n",
+    "    .sort_values(by = ['CITY_COUNT_BY_VAX_LOT'], ascending = False)\n",
+    "    .to_excel('tmp/cityCountsByClickedBatchcodeTable_EX8679_Germany_sorted.xlsx'))"
+   ]
+  },
   {
    "attachments": {},
    "cell_type": "markdown",