Merge pull request #46 from cmmorrow/BUGFIX-Categorical_numeric_group…

…_name_fix

Bugfix categorical numeric group name fix

sci_analysis/data/categorical.py

@@ -67,12 +67,21 @@ def __init__(self, sequence=None, name=None, order=None, dropna=False):
             self._summary = sequence.summary
         else:
             self._name = name
+            self._values = pd.Series(sequence)
             try:
-                self._values = pd.Series(sequence).astype('category')
+                self._values.astype('category')
             except TypeError:
-                self._values = pd.Series(flatten(sequence)).astype('category')
+                self._values = pd.Series(flatten(sequence))
             except ValueError:
                 self._values = pd.Series([])
+            # Try to preserve the original dtype of the categories.
+            try:
+                if not any(self._values % 1):
+                    self._values = self._values.astype(int)
+            except TypeError:
+                pass
+            self._values = self._values.astype('category')
+
             if order is not None:
                 if not is_iterable(order):
                     order = [order]
@@ -91,6 +100,7 @@ def __init__(self, sequence=None, name=None, order=None, dropna=False):
                 'percents': (counts / counts.sum() * 100) if not all(counts == 0) else 0.0
             })
             self._summary['categories'] = self._summary.index.to_series()
+
             if order is not None:
                 self._summary.sort_index(level=self._order, inplace=True, axis=0, na_position='last')
             else:

sci_analysis/test/test_cat_statistics.py

@@ -193,6 +193,75 @@ def test_106_single_category(self):
         self.assertEqual(str(test), output)
         self.assertListEqual(test.results, [{'Frequency': 2, 'Category': 'c', 'Rank': 1, 'Percent': 100.0}])
 
+    def test_107_numeric_group_name(self):
+        input_array = [1., 2., 1., 3., 3., 4.]
+        output = """
+
+Overall Statistics
+------------------
+
+Total            =  6
+Number of Groups =  4
+
+
+Statistics
+----------
+
+Rank          Frequency     Percent       Category      
+--------------------------------------------------------
+1             2              33.3333      1             
+1             2              33.3333      3             
+2             1              16.6667      2             
+2             1              16.6667      4             """
+        exp = CategoricalStatistics(input_array, display=False)
+        self.assertEqual(str(exp), output)
+
+    def test_108_year_group_name(self):
+        input_array = [2015, 2016, 2017, 2018, 2019]
+        exp = CategoricalStatistics(input_array, display=False)
+        output = """
+
+Overall Statistics
+------------------
+
+Total            =  5
+Number of Groups =  5
+
+
+Statistics
+----------
+
+Rank          Frequency     Percent       Category      
+--------------------------------------------------------
+1             1              20.0000      2015          
+1             1              20.0000      2016          
+1             1              20.0000      2017          
+1             1              20.0000      2018          
+1             1              20.0000      2019          """
+        self.assertEqual(str(exp), output)
+
+    def test_109_float_group_name(self):
+        input_array = [.123, .456, .789]
+        exp = CategoricalStatistics(input_array, display=True)
+        output = """
+
+Overall Statistics
+------------------
+
+Total            =  3
+Number of Groups =  3
+
+
+Statistics
+----------
+
+Rank          Frequency     Percent       Category      
+--------------------------------------------------------
+1             1              33.3333       0.1230       
+1             1              33.3333       0.4560       
+1             1              33.3333       0.7890       """
+        self.assertEqual(str(exp), output)
+
 
 if __name__ == '__main__':
     unittest.main()