Support more input types for categorical data. (#7220)

* Support more input types for categorical data.

* Shorten the type name from "categorical" to "c".
* Tests for np/cp array and scipy csr/csc/coo.
* Specify the type for feature info.

tests/cpp/tree/test_tree_model.cc

@@ -285,7 +285,7 @@ void TestCategoricalTreeDump(std::string format, std::string sep) {
   pos = str.find(cond_str, pos + 1);
   ASSERT_NE(pos, std::string::npos);
 
-  fmap.PushBack(0, "feat_0", "categorical");
+  fmap.PushBack(0, "feat_0", "c");
   fmap.PushBack(1, "feat_1", "q");
   fmap.PushBack(2, "feat_2", "int");
 

tests/python-gpu/test_from_cudf.py

@@ -172,7 +172,7 @@ Arrow specification.'''
         _test_cudf_metainfo(xgb.DeviceQuantileDMatrix)
 
     @pytest.mark.skipif(**tm.no_cudf())
-    def test_categorical(self):
+    def test_cudf_categorical(self):
         import cudf
         _X, _y = tm.make_categorical(100, 30, 17, False)
         X = cudf.from_pandas(_X)
@@ -180,11 +180,11 @@ Arrow specification.'''
 
         Xy = xgb.DMatrix(X, y, enable_categorical=True)
         assert len(Xy.feature_types) == X.shape[1]
-        assert all(t == "categorical" for t in Xy.feature_types)
+        assert all(t == "c" for t in Xy.feature_types)
 
         Xy = xgb.DeviceQuantileDMatrix(X, y, enable_categorical=True)
         assert len(Xy.feature_types) == X.shape[1]
-        assert all(t == "categorical" for t in Xy.feature_types)
+        assert all(t == "c" for t in Xy.feature_types)
 
 
 @pytest.mark.skipif(**tm.no_cudf())

tests/python-gpu/test_from_cupy.py

@@ -169,6 +169,19 @@ Arrow specification.'''
         X = cp.random.random((n, 2))
         xgb.DMatrix(X.toDlpack())
 
+    @pytest.mark.skipif(**tm.no_cupy())
+    def test_cupy_categorical(self):
+        import cupy as cp
+        n_features = 10
+        X, y = tm.make_categorical(10, n_features, n_categories=4, onehot=False)
+        X = cp.asarray(X.values.astype(cp.float32))
+        y = cp.array(y)
+        feature_types = ['c'] * n_features
+
+        assert isinstance(X, cp.ndarray)
+        Xy = xgb.DMatrix(X, y, feature_types=feature_types)
+        np.testing.assert_equal(np.array(Xy.feature_types), np.array(feature_types))
+
     @pytest.mark.skipif(**tm.no_cupy())
     def test_dlpack_device_dmat(self):
         import cupy as cp

tests/python/test_dmatrix.py

@@ -339,3 +339,44 @@ class TestDMatrix:
         Xy = xgb.DMatrix(X, y)
         assert Xy.num_row() == 10
         assert Xy.num_col() == 10
+
+    @pytest.mark.skipif(**tm.no_pandas())
+    def test_np_categorical(self):
+        n_features = 10
+        X, y = tm.make_categorical(10, n_features, n_categories=4, onehot=False)
+        X = X.values.astype(np.float32)
+        feature_types = ['c'] * n_features
+
+        assert isinstance(X, np.ndarray)
+        Xy = xgb.DMatrix(X, y, feature_types=feature_types)
+        np.testing.assert_equal(np.array(Xy.feature_types), np.array(feature_types))
+
+    def test_scipy_categorical(self):
+        from scipy import sparse
+        n_features = 10
+        X, y = tm.make_categorical(10, n_features, n_categories=4, onehot=False)
+        X = X.values.astype(np.float32)
+        feature_types = ['c'] * n_features
+
+        X[1, 3] = np.NAN
+        X[2, 4] = np.NAN
+        X = sparse.csr_matrix(X)
+
+        Xy = xgb.DMatrix(X, y, feature_types=feature_types)
+        np.testing.assert_equal(np.array(Xy.feature_types), np.array(feature_types))
+
+        X = sparse.csc_matrix(X)
+
+        Xy = xgb.DMatrix(X, y, feature_types=feature_types)
+        np.testing.assert_equal(np.array(Xy.feature_types), np.array(feature_types))
+
+        X = sparse.coo_matrix(X)
+
+        Xy = xgb.DMatrix(X, y, feature_types=feature_types)
+        np.testing.assert_equal(np.array(Xy.feature_types), np.array(feature_types))
+
+    def test_uri_categorical(self):
+        path = os.path.join(dpath, 'agaricus.txt.train')
+        feature_types = ["q"] * 5 + ["c"] + ["q"] * 120
+        Xy = xgb.DMatrix(path + "?indexing_mode=1", feature_types=feature_types)
+        np.testing.assert_equal(np.array(Xy.feature_types), np.array(feature_types))

tests/python/test_with_pandas.py

@@ -128,7 +128,7 @@ class TestPandas:
         X = pd.DataFrame({'f0': X})
         y = rng.randn(rows)
         m = xgb.DMatrix(X, y, enable_categorical=True)
-        assert m.feature_types[0] == 'categorical'
+        assert m.feature_types[0] == 'c'
 
     def test_pandas_sparse(self):
         import pandas as pd