[ENH] owheatmap: Add Split By combo box

Orange/widgets/visualize/owheatmap.py

@@ -1,7 +1,7 @@
 import math
 import itertools
 
-from collections import defaultdict, namedtuple
+from collections import namedtuple
 from types import SimpleNamespace as namespace
 
 import numpy as np
@@ -11,7 +11,7 @@
     QSizePolicy, QGraphicsScene, QGraphicsView, QGraphicsRectItem,
     QGraphicsWidget, QGraphicsSimpleTextItem, QGraphicsPixmapItem,
     QGraphicsGridLayout, QGraphicsLinearLayout, QGraphicsLayoutItem,
-    QFormLayout, QApplication
+    QFormLayout, QApplication, QComboBox
 )
 from AnyQt.QtGui import (
     QFontMetrics, QPen, QPixmap, QColor, QLinearGradient, QPainter,
@@ -24,12 +24,14 @@
 )
 import pyqtgraph as pg
 
+from orangewidget.utils.combobox import ComboBox
+
 from Orange.data import Domain, Table
 from Orange.data.sql.table import SqlTable
 import Orange.distance
 
 from Orange.clustering import hierarchical, kmeans
-from Orange.widgets.utils.itemmodels import DomainModel
+from Orange.widgets.utils.itemmodels import DomainModel, VariableListModel
 from Orange.widgets.utils.stickygraphicsview import StickyGraphicsView
 from Orange.widgets.utils import colorbrewer
 from Orange.widgets.utils.annotated_data import (create_annotated_table,
@@ -41,108 +43,11 @@
 from Orange.widgets.widget import Msg, Input, Output
 
 
-def split_domain(domain, split_label):
-    """Split the domain based on values of `split_label` value.
-    """
-    groups = defaultdict(list)
-    for attr in domain.attributes:
-        groups[attr.attributes.get(split_label)].append(attr)
-
-    attr_values = [attr.attributes.get(split_label)
-                   for attr in domain.attributes]
-
-    domains = []
-    for value, attrs in groups.items():
-        group_domain = Domain(attrs, domain.class_vars, domain.metas)
-
-        domains.append((value, group_domain))
-
-    if domains:
-        assert all(len(dom) == len(domains[0][1]) for _, dom in domains)
-
-    return sorted(domains, key=lambda t: attr_values.index(t[0]))
-
-
-def vstack_by_subdomain(data, sub_domains):
-    domain = sub_domains[0]
-    newtable = Table(domain)
-
-    for sub_dom in sub_domains:
-        sub_data = data.transform(sub_dom)
-        # TODO: improve O(N ** 2)
-        newtable.extend(sub_data)
-
-    return newtable
-
-
-def select_by_class(data, class_):
-    indices = select_by_class_indices(data, class_)
-    return data[indices]
-
-
-def select_by_class_indices(data, class_):
-    col, _ = data.get_column_view(data.domain.class_var)
-    return col == class_
-
-
-def group_by_unordered(iterable, key):
-    groups = defaultdict(list)
-    for item in iterable:
-        groups[key(item)].append(item)
-    return groups.items()
-
-
-def barycenter(a, axis=0):
-    assert 0 <= axis < 2
-    a = np.asarray(a)
-    N = a.shape[axis]
-    tileshape = [1 if i != axis else a.shape[i] for i in range(a.ndim)]
-    xshape = list(a.shape)
-    xshape[axis] = 1
-    X = np.tile(np.reshape(np.arange(N), tileshape), xshape)
-    amin = np.nanmin(a, axis=axis, keepdims=True)
-    weights = a - amin
-    weights[np.isnan(weights)] = 0
-    wsum = np.sum(weights, axis=axis)
-    mask = wsum <= np.finfo(float).eps
-    if axis == 1:
-        weights[mask, :] = 1
-    else:
-        weights[:, mask] = 1
-
-    return np.average(X, weights=weights, axis=axis)
-
-
 def kmeans_compress(X, k=50):
     km = kmeans.KMeans(n_clusters=k, n_init=5, random_state=42)
     return km.get_model(X)
 
 
-def candidate_split_labels(data):
-    """
-    Return candidate labels on which we can split the data.
-    """
-    groups = defaultdict(list)
-    for attr in data.domain.attributes:
-        for item in attr.attributes.items():
-            groups[item].append(attr)
-
-    by_keys = defaultdict(list)
-    for (key, _), attrs in groups.items():
-        by_keys[key].append(attrs)
-
-    # Find the keys for which all values have the same number
-    # of attributes.
-    candidates = []
-    for key, groups in by_keys.items():
-        count = len(groups[0])
-        if all(len(attrs) == count for attrs in groups) and \
-                len(groups) > 1 and count > 1:
-            candidates.append(key)
-
-    return candidates
-
-
 def leaf_indices(tree):
     return [leaf.value.index for leaf in hierarchical.leaves(tree)]
 
@@ -384,9 +289,23 @@ def cluster_ord(self):
     [name for name, _, in _color_palettes].index("Blue-Yellow")
 
 
+def cbselect(cb: QComboBox, value, role: Qt.ItemDataRole = Qt.EditRole) -> None:
+    """
+    Find and select the `value` in the `cb` QComboBox.
+
+    Parameters
+    ----------
+    cb: QComboBox
+    value: Any
+    role: Qt.ItemDataRole
+        The data role in the combo box model to match value against
+    """
+    cb.setCurrentIndex(cb.findData(value, role))
+
+
 class OWHeatMap(widget.OWWidget):
     name = "Heat Map"
-    description = "Plot a heat map for a pair of attributes."
+    description = "Plot a data matrix heatmap."
     icon = "icons/Heatmap.svg"
     priority = 260
     keywords = []
@@ -418,6 +337,8 @@ class Outputs:
     legend = settings.Setting(True)
     # Annotations
     annotation_var = settings.ContextSetting(None)
+    split_by_var = settings.ContextSetting(None)
+
     # Stored color palette settings
     color_settings = settings.Setting(None)
     user_palettes = settings.Setting([])
@@ -542,6 +463,33 @@ def __init__(self):
             cluster_box, self, "row_clustering", "Rows",
             callback=self.update_clustering_examples)
 
+        box = gui.vBox(self.controlArea, "Split By")
+
+        self.row_split_model = DomainModel(
+            placeholder="(None)",
+            valid_types=(Orange.data.DiscreteVariable,),
+            parent=self,
+        )
+        self.row_split_cb = cb = ComboBox(
+            enabled=not self.merge_kmeans,
+            sizeAdjustPolicy=ComboBox.AdjustToMinimumContentsLengthWithIcon,
+            minimumContentsLength=14,
+            toolTip="Split the heatmap vertically by a categorical column"
+        )
+        self.row_split_cb.setModel(self.row_split_model)
+        self.connect_control(
+            "split_by_var", lambda value, cb=cb: cbselect(cb, value)
+        )
+        self.connect_control(
+            "merge_kmeans", self.row_split_cb.setDisabled
+        )
+        self.split_by_var = None
+
+        self.row_split_cb.activated.connect(
+            self.__on_split_rows_activated
+        )
+        box.layout().addWidget(self.row_split_cb)
+
         box = gui.vBox(self.controlArea, 'Annotation && Legends')
 
         gui.checkBox(box, self, 'legend', 'Show legend',
@@ -626,6 +574,8 @@ def clear(self):
         self.merge_indices = None
         self.annotation_model.set_domain(None)
         self.annotation_var = None
+        self.row_split_model.set_domain(None)
+        self.split_by_var = None
         self.clear_scene()
         self.selected_rows = []
         self.__columns_cache.clear()
@@ -705,7 +655,14 @@ def set_dataset(self, data=None):
         if data is not None:
             self.annotation_model.set_domain(self.input_data.domain)
             self.annotation_var = None
+            self.row_split_model.set_domain(data.domain)
+            if data.domain.has_discrete_class:
+                self.split_by_var = data.domain.class_var
+            else:
+                self.split_by_var = None
             self.openContext(self.input_data)
+            if self.split_by_var not in self.row_split_model:
+                self.split_by_var = None
 
         self.update_heatmaps()
         if data is not None and self.__pending_selection is not None:
@@ -715,6 +672,14 @@ def set_dataset(self, data=None):
 
         self.unconditional_commit()
 
+    def __on_split_rows_activated(self):
+        self.set_split_variable(self.row_split_cb.currentData(Qt.EditRole))
+
+    def set_split_variable(self, var):
+        if var != self.split_by_var:
+            self.split_by_var = var
+            self.update_heatmaps()
+
     def update_heatmaps(self):
         if self.data is not None:
             self.clear_scene()
@@ -727,7 +692,9 @@ def update_heatmaps(self):
             elif self.merge_kmeans and len(self.data) < 3:
                 self.Error.not_enough_instances_k_means()
             else:
-                self.construct_heatmaps(self.data)
+                self.heatmapparts = self.construct_heatmaps(
+                    self.data, self.split_by_var
+                )
                 self.construct_heatmaps_scene(
                     self.heatmapparts, self.effective_data)
                 self.selected_rows = []
@@ -741,7 +708,7 @@ def update_merge(self):
             self.update_heatmaps()
             self.commit()
 
-    def _make_parts(self, data, group_var=None, group_key=None):
+    def _make_parts(self, data, group_var=None):
         """
         Make initial `Parts` for data, split by group_var, group_key
         """
@@ -758,20 +725,11 @@ def _make_parts(self, data, group_var=None, group_key=None):
                                   sortindices=None,
                                   cluster=None, cluster_ordered=None)]
 
-        if group_key is not None:
-            col_groups = split_domain(data.domain, group_key)
-            assert len(col_groups) > 0
-            col_indices = [np.array([data.domain.index(var) for var in group])
-                           for _, group in col_groups]
-            col_groups = [ColumnPart(title=name, domain=d, indices=ind,
-                                     cluster=None, cluster_ordered=None)
-                          for (name, d), ind in zip(col_groups, col_indices)]
-        else:
-            col_groups = [
-                ColumnPart(
-                    title=None, indices=slice(0, len(data.domain.attributes)),
-                    domain=data.domain, cluster=None, cluster_ordered=None)
-            ]
+        col_groups = [
+            ColumnPart(
+                title=None, indices=slice(0, len(data.domain.attributes)),
+                domain=data.domain, cluster=None, cluster_ordered=None)
+        ]
 
         minv, maxv = np.nanmin(data.X), np.nanmax(data.X)
         return Parts(row_groups, col_groups, span=(minv, maxv))
@@ -806,11 +764,10 @@ def cluster_rows(self, data, parts):
 
             row_groups.append(row._replace(cluster=cluster, cluster_ordered=cluster_ord))
 
-        return parts._replace(columns=parts.columns, rows=row_groups)
+        return parts._replace(rows=row_groups)
 
     def cluster_columns(self, data, parts):
-        if len(parts.columns) > 1:
-            data = vstack_by_subdomain(data, [col.domain for col in parts.columns])
+        assert len(parts.columns) == 1, "columns split is no longer supported"
         assert all(var.is_continuous for var in data.domain.attributes)
 
         col0 = parts.columns[0]
@@ -839,21 +796,9 @@ def cluster_columns(self, data, parts):
 
         col_groups = [col._replace(cluster=cluster, cluster_ordered=cluster_ord)
                       for col in parts.columns]
-        return parts._replace(columns=col_groups, rows=parts.rows)
+        return parts._replace(columns=col_groups)
 
-    def construct_heatmaps(self, data, split_label=None):
-        if split_label is not None:
-            groups = split_domain(data.domain, split_label)
-            assert len(groups) > 0
-        else:
-            groups = [("", data.domain)]
-
-        if data.domain.has_discrete_class:
-            group_var = data.domain.class_var
-        else:
-            group_var = None
-
-        group_label = split_label
+    def construct_heatmaps(self, data, group_var=None) -> 'Parts':
         if self.merge_kmeans:
             if self.kmeans_model is None:
                 effective_data = self.input_data.transform(
@@ -890,18 +835,14 @@ def construct_heatmaps(self, data, split_label=None):
 
         self.__update_clustering_enable_state(effective_data)
 
-        parts = self._make_parts(effective_data, group_var, group_label)
+        parts = self._make_parts(effective_data, group_var)
         # Restore/update the row/columns items descriptions from cache if
         # available
         rows_cache_key = (group_var,
                           self.merge_kmeans_k if self.merge_kmeans else None)
         if rows_cache_key in self.__rows_cache:
             parts = parts._replace(rows=self.__rows_cache[rows_cache_key].rows)
 
-        if group_label in self.__columns_cache:
-            parts = parts._replace(
-                columns=self.__columns_cache[group_label].columns)
-
         if self.row_clustering:
             assert len(effective_data) <= OWHeatMap._MaxOrderedClustering
             parts = self.cluster_rows(effective_data, parts)
@@ -913,9 +854,7 @@ def construct_heatmaps(self, data, split_label=None):
 
         # Cache the updated parts
         self.__rows_cache[rows_cache_key] = parts
-        self.__columns_cache[group_label] = parts
-
-        self.heatmapparts = parts
+        return parts
 
     def construct_heatmaps_scene(self, parts, data):
         def select_row(item):
@@ -1521,8 +1460,10 @@ def send_report(self):
         self.report_items((
             ("Columns:", "Clustering" if self.col_clustering else "No sorting"),
             ("Rows:", "Clustering" if self.row_clustering else "No sorting"),
+            ("Split:",
+             self.split_by_var is not None and self.split_by_var.name),
             ("Row annotation",
-             self.annotation_var is not None and self.annotation_var.name)
+             self.annotation_var is not None and self.annotation_var.name),
         ))
         self.report_plot()
 

Orange/widgets/visualize/tests/test_owheatmap.py

@@ -198,6 +198,17 @@ def test_saved_selection(self):
         self.send_signal(w.Inputs.data, iris, widget=w)
         self.assertEqual(len(self.get_output(w.Outputs.selected_data)), 21)
 
+    def test_set_split_var(self):
+        data = Table("brown-selected")
+        w = self.widget
+        self.send_signal(self.widget.Inputs.data, data, widget=w)
+        self.assertIs(w.split_by_var, data.domain.class_var)
+        self.assertEqual(len(w.heatmapparts.rows),
+                         len(data.domain.class_var.values))
+        w.set_split_variable(None)
+        self.assertIs(w.split_by_var, None)
+        self.assertEqual(len(w.heatmapparts.rows), 1)
+
 
 if __name__ == "__main__":
     unittest.main()