pvals measures for subtotals

src/cr/cube/matrix.py

@@ -950,6 +950,10 @@ def table_margin(self):
     def values(self):
         return np.sum(np.array([row.values for row in self._addend_vectors]), axis=0)
 
+    @lazyproperty
+    def _pvals(self):
+        return 2 * (1 - norm.cdf(np.abs(self._zscore)))
+
     @lazyproperty
     def _zscore(self):
         variance = (
@@ -979,6 +983,14 @@ def pruned(self):
             np.array([row.base for row in self._matrix.rows])
         )
 
+    @lazyproperty
+    def pvals(self):
+        return self._pvals
+
+    @lazyproperty
+    def zscore(self):
+        return self._zscore
+
     @lazyproperty
     def _addend_vectors(self):
         return tuple(
@@ -999,10 +1011,6 @@ def _expected_counts(self):
     def _opposite_margins(self):
         return self._addend_vectors[0].opposite_margins
 
-    @lazyproperty
-    def zscore(self):
-        return self._zscore
-
 
 class _InsertionRow(_BaseMatrixInsertionVector):
     """Represents an inserted (subtotal) row."""
@@ -1021,7 +1029,7 @@ def pruned(self):
 
     @lazyproperty
     def pvals(self):
-        return np.array([np.nan] * len(self._matrix.columns))
+        return self._pvals
 
     @lazyproperty
     def zscore(self):
@@ -1141,11 +1149,7 @@ def proportions(self):
 
     @lazyproperty
     def pvals(self):
-        return np.array(
-            tuple([np.nan] * len(self._top_values))
-            + self._interleaved_pvals
-            + tuple([np.nan] * len(self._bottom_values))
-        )
+        return np.array(self._top_pvals + self._interleaved_pvals + self._bottom_pvals)
 
     @lazyproperty
     def table_proportions(self):
@@ -1178,6 +1182,12 @@ def _bottom_base_values(self):
             for col in self._bottom_insertions
         )
 
+    @lazyproperty
+    def _bottom_pvals(self):
+        return tuple(
+            vector.pvals[self._vector_idx] for vector in self._bottom_insertions
+        )
+
     @lazyproperty
     def _bottom_values(self):
         return tuple(
@@ -1235,7 +1245,7 @@ def _interleaved_pvals(self):
             pvals.append(value)
             for inserted_vector in self._opposite_inserted_vectors:
                 if i == inserted_vector.anchor:
-                    pvals.append(np.nan)
+                    pvals.append(inserted_vector.pvals[self._vector_idx])
         return tuple(pvals)
 
     @lazyproperty
@@ -1261,6 +1271,13 @@ def _interleaved_zscore(self):
                     zscore.append(inserted_vector.zscore[self._vector_idx])
         return tuple(zscore)
 
+    @lazyproperty
+    def _top_base_values(self):
+        return tuple(
+            np.sum(self._base_vector.base_values[col.addend_idxs])
+            for col in self._top_insertions
+        )
+
     @lazyproperty
     def _top_insertions(self):
         return tuple(
@@ -1270,15 +1287,8 @@ def _top_insertions(self):
         )
 
     @lazyproperty
-    def _top_zscores(self):
-        return tuple(vector.zscore[self._vector_idx] for vector in self._top_insertions)
-
-    @lazyproperty
-    def _top_base_values(self):
-        return tuple(
-            np.sum(self._base_vector.base_values[col.addend_idxs])
-            for col in self._top_insertions
-        )
+    def _top_pvals(self):
+        return tuple(vector.pvals[self._vector_idx] for vector in self._top_insertions)
 
     @lazyproperty
     def _top_values(self):
@@ -1287,6 +1297,10 @@ def _top_values(self):
             for col in self._top_insertions
         )
 
+    @lazyproperty
+    def _top_zscores(self):
+        return tuple(vector.zscore[self._vector_idx] for vector in self._top_insertions)
+
 
 class _VectorAfterHiding(_BaseTransformationVector):
     """Reflects a row or column with hidden elements removed."""
@@ -1379,14 +1393,6 @@ def column_index(self):
     def label(self):
         return self._base_vector.label
 
-    @lazyproperty
-    def _opposing_order(self):
-        return (
-            slice(None)
-            if self._opposing_order_arg is None
-            else self._opposing_order_arg
-        )
-
     @lazyproperty
     def pvals(self):
         return self._base_vector.pvals
@@ -1399,6 +1405,14 @@ def values(self):
     def zscore(self):
         return self._base_vector.zscore
 
+    @lazyproperty
+    def _opposing_order(self):
+        return (
+            slice(None)
+            if self._opposing_order_arg is None
+            else self._opposing_order_arg
+        )
+
 
 # ===OPERAND VECTORS===
 
@@ -1513,14 +1527,6 @@ def table_margin(self):
     def values(self):
         return self._counts
 
-    @lazyproperty
-    def _residuals(self):
-        return self.values - self._expected_counts
-
-    @lazyproperty
-    def _expected_counts(self):
-        return self._opposite_margins * self.margin / self.table_margin
-
     @lazyproperty
     def zscore(self):
         variance = (
@@ -1535,6 +1541,14 @@ def zscore(self):
 
         return self._residuals / np.sqrt(variance)
 
+    @lazyproperty
+    def _residuals(self):
+        return self.values - self._expected_counts
+
+    @lazyproperty
+    def _expected_counts(self):
+        return self._opposite_margins * self.margin / self.table_margin
+
 
 class _CatXMrVector(_CategoricalVector):
     """Used for categorical dimension when opposing dimension is multiple-response."""

tests/integration/test_cubepart.py

@@ -89,13 +89,13 @@ def it_calculates_various_measures(self):
         np.testing.assert_almost_equal(
             slice_.pvals,
             [
-                [0.03851757, np.nan, 0.54097586, 0.21071341, np.nan, 0.23299113],
+                [0.03851757, 0.0922145, 0.54097586, 0.21071341, np.nan, 0.23299113],
+                [0.04198008, 0.0922145, 0.50508577, 0.28105398, np.nan, 0.1797169],
+                [0.73113976, 0.41072494, 0.28019785, 0.32642279, np.nan, 0.79310382],
+                [0.36684711, 0.29203707, 0.98652895, 0.85178994, np.nan, 0.00305394],
+                [0.06398878, 0.47430453, 0.21130996, 0.26884987, np.nan, 0.4212984],
                 [np.nan, np.nan, np.nan, np.nan, np.nan, np.nan],
-                [0.73113976, np.nan, 0.28019785, 0.32642279, np.nan, 0.79310382],
-                [0.36684711, np.nan, 0.98652895, 0.85178994, np.nan, 0.00305394],
-                [0.06398878, np.nan, 0.21130996, 0.26884987, np.nan, 0.4212984],
-                [np.nan, np.nan, np.nan, np.nan, np.nan, np.nan],
-                [0.82020207, np.nan, 0.91486794, 0.58880283, np.nan, 0.75048675],
+                [0.82020207, 0.70318269, 0.91486794, 0.58880283, np.nan, 0.75048675],
             ],
         )