def test_construct_variables_discrete(self):
data = Table("iris")
name = "Discrete Variable"
expression = ("iris_one if iris == 'Iris-setosa' else iris_two "
"if iris == 'Iris-versicolor' else iris_three")
values = ("iris one", "iris two", "iris three")
desc = PyListModel([
DiscreteDescriptor(
name=name,
expression=expression,
values=values,
base_value=-1,
ordered=False,
)
])
data = Table(
Domain(
list(data.domain.attributes) +
construct_variables(desc, data.domain.variables),
data.domain.class_vars,
data.domain.metas,
),
data,
)
self.assertTrue(isinstance(data.domain[name], DiscreteVariable))
self.assertEqual(data.domain[name].values, list(values))
for i in range(3):
self.assertEqual(data[i * 50, name], values[i])
def test_construct_numeric_names(self):
data = Table("iris")
data.domain.attributes[0].name = "0.1"
data.domain.attributes[1].name = "1"
desc = PyListModel([
ContinuousDescriptor(name="S",
expression="_0_1 + _1",
number_of_decimals=3)
])
nv = construct_variables(desc, data.domain)
ndata = Table(Domain(nv, None), data)
np.testing.assert_array_equal(ndata.X[:, 0], data.X[:, :2].sum(axis=1))
ContinuousVariable._clear_all_caches()
def test_construct_variables_string(self):
data = Table("iris")
name = 'String Variable'
expression = "str(iris) + '_name'"
desc = PyListModel(
[StringDescriptor(name=name, expression=expression)])
data = data.transform(
Domain(data.domain.attributes, data.domain.class_vars,
list(data.domain.metas) + construct_variables(desc, data)))
self.assertTrue(isinstance(data.domain[name], StringVariable))
for i in range(3):
self.assertEqual(data[i * 50, name],
str(data[i * 50, "iris"]) + "_name")
def test_construct_numeric_names():
data = Table("iris")
newdomain = Domain(
(ContinuousVariable("0.1"), ContinuousVariable("1")) +
data.domain.attributes[2:], data.domain.class_var)
data = Table.from_numpy(newdomain, data.X, data.Y)
desc = PyListModel([
ContinuousDescriptor(name="S",
expression="_0_1 + _1",
number_of_decimals=3)
])
nv = construct_variables(desc, data)
ndata = Table(Domain(nv, None), data)
np.testing.assert_array_equal(ndata.X[:, 0], data.X[:, :2].sum(axis=1))
def test_unicode_normalization():
micro = "\u00b5"
domain = Domain([ContinuousVariable(micro)])
name = 'Micro Variable'
expression = micro
desc = PyListModel(
[ContinuousDescriptor(name=name, expression=expression,
number_of_decimals=2)]
)
data = Table.from_numpy(domain, np.arange(5).reshape(5, 1))
data = data.transform(Domain(data.domain.attributes,
[],
construct_variables(desc, data)))
np.testing.assert_equal(data.X, data.metas)
def test_construct_numeric_names(self):
data = Table("iris")
data.domain.attributes[0].name = "0.1"
data.domain.attributes[1].name = "1"
desc = PyListModel(
[ContinuousDescriptor(name="S",
expression="_0_1 + _1",
number_of_decimals=3)]
)
nv = construct_variables(desc, data.domain.variables)
ndata = Table(Domain(nv, None), data)
np.testing.assert_array_equal(ndata.X[:, 0],
data.X[:, :2].sum(axis=1))
ContinuousVariable._clear_all_caches()
def test_construct_variables_datetime(self):
data = Table("housing")
name = 'Date'
expression = '"2019-07-{:02}".format(int(MEDV/3))'
featuremodel = PyListModel(
[DateTimeDescriptor(name=name, expression=expression)])
data = Table(
Domain(
list(data.domain.attributes) +
construct_variables(featuremodel, data),
data.domain.class_vars, data.domain.metas), data)
self.assertTrue(isinstance(data.domain[name], TimeVariable))
for row in data:
self.assertEqual("2019-07-{:02}".format(int(row["MEDV"] / 3)),
str(row["Date"])[:10])
def test_construct_variables_string(self):
data = Table("iris")
name = 'String Variable'
expression = "str(iris) + '_name'"
desc = PyListModel(
[StringDescriptor(name=name, expression=expression)]
)
data = Table(Domain(data.domain.attributes,
data.domain.class_vars,
list(data.domain.metas) +
construct_variables(desc, data.domain)),
data)
self.assertTrue(isinstance(data.domain[name], StringVariable))
for i in range(3):
self.assertEqual(data[i * 50, name],
str(data[i * 50, "iris"]) + "_name")
def test_construct_variables_continuous(self):
data = Table("iris")
name = 'Continuous Variable'
expression = "pow(sepal_length + sepal_width, 2)"
featuremodel = PyListModel(
[ContinuousDescriptor(name=name, expression=expression,
number_of_decimals=2)]
)
data = Table(Domain(list(data.domain.attributes) +
construct_variables(featuremodel, data.domain),
data.domain.class_vars,
data.domain.metas), data)
self.assertTrue(isinstance(data.domain[name], ContinuousVariable))
for i in range(3):
self.assertEqual(data[i * 50, name],
pow(data[i * 50, 0] + data[i * 50, 1], 2))
def test_construct_variables_continuous(self):
data = Table("iris")
name = 'Continuous Variable'
expression = "pow(sepal_length + sepal_width, 2)"
featuremodel = PyListModel(
[ContinuousDescriptor(name=name, expression=expression,
number_of_decimals=2)]
)
data = Table(Domain(list(data.domain.attributes) +
construct_variables(featuremodel, data.domain),
data.domain.class_vars,
data.domain.metas), data)
self.assertTrue(isinstance(data.domain[name], ContinuousVariable))
for i in range(3):
self.assertEqual(data[i * 50, name],
pow(data[i * 50, 0] + data[i * 50, 1], 2))
def test_construct_variables_discrete(self):
data = Table("iris")
name = 'Discrete Variable'
expression = "iris_one if iris == 'Iris-setosa' else iris_two " \
"if iris == 'Iris-versicolor' else iris_three"
values = ('iris one', 'iris two', 'iris three')
desc = PyListModel(
[DiscreteDescriptor(name=name, expression=expression,
values=values, base_value=-1, ordered=False)]
)
data = Table(Domain(list(data.domain.attributes) +
construct_variables(desc, data.domain),
data.domain.class_vars,
data.domain.metas), data)
self.assertTrue(isinstance(data.domain[name], DiscreteVariable))
self.assertEqual(data.domain[name].values, list(values))
for i in range(3):
self.assertEqual(data[i * 50, name], values[i])
def test_construct_variables_discrete(self):
data = Table("iris")
name = 'Discrete Variable'
expression = "iris_one if iris == 'Iris-setosa' else iris_two " \
"if iris == 'Iris-versicolor' else iris_three"
values = ('iris one', 'iris two', 'iris three')
desc = PyListModel(
[DiscreteDescriptor(name=name, expression=expression,
values=values, ordered=False)]
)
data = data.transform(Domain(list(data.domain.attributes) +
construct_variables(desc, data),
data.domain.class_vars,
data.domain.metas))
self.assertTrue(isinstance(data.domain[name], DiscreteVariable))
self.assertEqual(data.domain[name].values, values)
for i in range(3):
self.assertEqual(data[i * 50, name], values[i])
def test_construct_variables_discrete_no_values(self):
data = Table("iris")
name = 'Discrete Variable'
expression = "str(iris)[-1]" # last letter - a or r
values = ()
desc = PyListModel(
[DiscreteDescriptor(name=name, expression=expression,
values=values, ordered=False)]
)
data = data.transform(Domain(list(data.domain.attributes) +
construct_variables(desc, data),
data.domain.class_vars,
data.domain.metas))
newvar = data.domain[name]
self.assertTrue(isinstance(newvar, DiscreteVariable))
self.assertEqual(set(data.domain[name].values), set("ar"))
for i in range(3):
inst = data[i * 50]
self.assertEqual(str(inst[name]), str(inst["iris"])[-1])
