Merge 5a825b1 into 86db076

sci_analysis/analysis/comparison.py

@@ -222,9 +222,6 @@ def __init__(self, xdata, ydata=None, groups=None, alpha=None, display=True):
             vector = Vector(xdata, other=ydata, groups=groups)
         if vector.is_empty():
             raise NoDataError("Cannot perform test because there is no data")
-        for grp, seq in vector.groups.items():
-            if len(seq) <= self._min_size:
-                raise MinimumSizeError("length of {} is less than the minimum size {}".format(grp, self._min_size))
         super(GroupComparison, self).__init__(vector, display=display)
         self._alpha = alpha or self._default_alpha
         self.logic()
@@ -251,6 +248,9 @@ def __init__(self, xdata, ydata=None, groups=None, alpha=None, display=True):
 
     def run(self):
         out = []
+        # Remove any groups that are less than or equal to the minimum value from analysis.
+        small_grps = [grp for grp, seq in self.data.groups.items() if len(seq) <= self._min_size]
+        self.data.drop_groups(small_grps)
         if NormTest(*self.data.flatten(), display=False, alpha=self._alpha).p_value > self._alpha:
             r = "pearson"
             func = pearsonr
@@ -308,6 +308,9 @@ class GroupLinearRegression(GroupComparison):
 
     def run(self):
         out = []
+        # Remove any groups that are less than or equal to the minimum value from analysis.
+        small_grps = [grp for grp, seq in self.data.groups.items() if len(seq) <= self._min_size]
+        self.data.drop_groups(small_grps)
         for grp, pairs in self.data.paired_groups.items():
             slope, intercept, r, p_value, std_err = linregress(*pairs)
             count = len(pairs[0])
@@ -321,6 +324,8 @@ def run(self):
                 self._p_value: p_value,
                 self._group_name: str(grp)
             })
+        if not out:
+            raise NoDataError
         self._results = DataFrame(out).sort_values(self._group_name).to_dict(orient='records')
 
     def __str__(self):

sci_analysis/data/numeric.py

@@ -4,7 +4,7 @@
 
 # Import from local
 from .data import Data, is_data
-from .data_operations import flatten
+from .data_operations import flatten, is_iterable
 
 
 class EmptyVectorError(Exception):
@@ -143,7 +143,7 @@ def drop_nan(self):
 
         Returns
         -------
-        arr : pandas.DataFrame
+        arr : pandas.Series
             A copy of the Numeric object's internal Series with all NaN values removed.
         """
         return self._values.dropna(how='any', subset=[self._ind])
@@ -156,10 +156,31 @@ def drop_nan_intersect(self):
         Returns
         -------
         arr : pandas.DataFrame
-            A tuple of numpy Arrays corresponding to the internal Vector and seq with all nan values removed.
+            A copy of the Numeric object's internal DataFrame with all nan values removed.
         """
         return self._values.dropna(how='any', subset=[self._ind, self._dep])
 
+    def drop_groups(self, grps):
+        """Drop the specified group name from the Numeric object.
+
+        Parameters
+        ----------
+        grps : str|int|list[str]|list[int]
+            The name of the group to remove.
+
+        Returns
+        -------
+        arr : pandas.DataFrame
+            A copy of the Numeric object's internal DataFrame with all records belonging to the specified group removed.
+
+        """
+        if not is_iterable(grps):
+            grps = [grps]
+        dropped = self._values.query("{} not in {}".format(self._grp, grps)).copy()
+        dropped[self._grp] = dropped[self._grp].cat.remove_categories(grps)
+        self._values = dropped
+        return dropped
+
     @property
     def data_type(self):
         return self._type

sci_analysis/test/test_analyze.py

@@ -457,6 +457,22 @@ def test_140_scatter_highlight_labels(self):
             ), ['Bivariate']
         )
 
+    def test_141_scatter_groups_one_below_min_size(self):
+        np.random.seed(self._seed)
+        df = pd.DataFrame(np.random.randn(100, 2), columns=list('xy'))
+        df['groups'] = np.random.choice(list('ABC'), len(df)).tolist()
+        df.at[24, 'groups'] = "D"
+        self.assertEqual(
+            analyze(
+                df['x'],
+                df['y'],
+                df['groups'],
+                debug=True,
+                save_to='{}test_analyze_141'.format(self.save_path)
+            ),
+            ['Group Bivariate']
+        )
+
 
 if __name__ == '__main__':
     unittest.main()