starting BarChartDescriptionTablesTest

src/BarChartDescriptionTables.py

@@ -0,0 +1,12 @@
+import math
+
+class BarChartDescriptionTables:
+
+    @staticmethod
+    def filterValidJensenShannonDistances(barChartDescriptionTable):
+        return barChartDescriptionTable[barChartDescriptionTable.apply(BarChartDescriptionTables._isValidJensenShannonDistance, axis = 'columns')]
+
+    @staticmethod
+    def _isValidJensenShannonDistance(barChartDescription):
+        jensenShannonDistance = barChartDescription['BAR_CHART_DESCRIPTION']['Jensen-Shannon distance']
+        return not math.isnan(jensenShannonDistance)

src/BarChartDescriptionTablesTest.py

@@ -0,0 +1,62 @@
+import unittest
+import pandas as pd
+import numpy as np
+from pandas.testing import assert_frame_equal
+from TestHelper import TestHelper
+from BarChartDescriptionTables import BarChartDescriptionTables
+
+class BarChartDescriptionTablesTest(unittest.TestCase):
+
+    def test_filterValidJensenShannonDistances(self):
+        # Given
+        barChartDescriptionTable = TestHelper.createDataFrame(
+                columns = ['BAR_CHART_DESCRIPTION'],
+                data = [
+                        [
+                            {
+                                'countries':                        ['Germany', 'Hungary'],
+                                'Adverse Reaction Reports guessed': [0,         0],
+                                'Adverse Reaction Reports known':   [20,        30],
+                                'Jensen-Shannon distance':          np.nan
+                            }
+                        ],
+                        [
+                            {
+                                'countries':                        ['Germany'],
+                                'Adverse Reaction Reports guessed': [70],
+                                'Adverse Reaction Reports known':   [80],
+                                'Jensen-Shannon distance':          0.4711
+                            }
+                        ]
+                       ],
+                index = pd.Index(
+                    [
+                        '!D0181',
+                        '# 009C01A'
+                    ],
+                    name = 'VAX_LOT'))
+        
+        # When
+        barChartDescriptionTableResult = BarChartDescriptionTables.filterValidJensenShannonDistances(barChartDescriptionTable)
+
+        # Then
+        assert_frame_equal(
+            barChartDescriptionTableResult,
+            TestHelper.createDataFrame(
+                columns = ['BAR_CHART_DESCRIPTION'],
+                data = [
+                        [
+                            {
+                                'countries':                        ['Germany'],
+                                'Adverse Reaction Reports guessed': [70],
+                                'Adverse Reaction Reports known':   [80],
+                                'Jensen-Shannon distance':          0.4711
+                            }
+                        ]
+                       ],
+                index = pd.Index(
+                    [
+                        '# 009C01A',
+                    ],
+                    name = 'VAX_LOT')),
+            check_dtype = True)

src/HowBadIsMyBatch.ipynb

@@ -25,6 +25,7 @@
     "from IOUtils import IOUtils\n",
     "import os\n",
     "import pandas as pd\n",
+    "from BarChartDescriptionTables import BarChartDescriptionTables\n",
     "\n",
     "pd.set_option('display.max_rows', 100)\n",
     "pd.set_option('display.max_columns', None)\n"
@@ -220,8 +221,41 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "barChartDescriptionTableWithJensenShannonDistance = JensenShannonDistance2BarChartDescriptionColumnAdder.addJensenShannonDistance2BarChartDescriptionColumn(barChartDescriptionTable)\n",
-    "barChartDescriptionTableWithJensenShannonDistance"
+    "barChartDescriptionTable = JensenShannonDistance2BarChartDescriptionColumnAdder.addJensenShannonDistance2BarChartDescriptionColumn(barChartDescriptionTable)\n",
+    "barChartDescriptionTable"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "39c83928",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "barChartDescriptionTable = BarChartDescriptionTables.filterValidJensenShannonDistances(barChartDescriptionTable)\n",
+    "barChartDescriptionTable"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "123be407",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "batchcodes2Retain = barChartDescriptionTable.index.values\n",
+    "batchCodeTable = batchCodeTable[batchCodeTable['Batch'].isin(batchcodes2Retain)]\n",
+    "batchCodeTable"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "3bd5d78d",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "IOUtils.saveDataFrameAsJson(batchCodeTable, '../docs/data/batchCodeTables/Global.json')"
    ]
   },
   {
@@ -232,7 +266,7 @@
    "outputs": [],
    "source": [
     "IOUtils.saveDictAsJson(\n",
-    "    BarChartDescriptionTable2DictionaryConverter.convert2Dictionary(barChartDescriptionTableWithJensenShannonDistance, internationalVaersCovid19),\n",
+    "    BarChartDescriptionTable2DictionaryConverter.convert2Dictionary(barChartDescriptionTable, internationalVaersCovid19),\n",
     "    '../docs/data/barChartDescriptionTable.json')\n"
    ]
   },

src/help.txt

@@ -5,6 +5,9 @@ FK-FIXME:
 FK-TODO:
 - add google captcha to batchCodeTable.html
 - Symptomhistogramm
+- Filter für ein Barchart:
+  - im Barchart gibt es mindestens N Guessed-Einträge
+  - im Barchart gibt es mindestens ein Country mit Guessed > Known
 
 anacron job:
 sudo cp src/intensivstationen_howbadismybatch.sh /etc/cron.daily/intensivstationen_howbadismybatch