def test_evalenv_equality():
a = 1
b = "hello friend"
c = [1, 2, 3]
d = {"a": 1, "b": 2}
# The comparison is made in terms of the ids of the namespaces
env = Environment.capture()
env2 = Environment.capture()
assert env == env2
env_outer = Environment.capture(1)
assert env != env_outer
def test_evalenv_capture():
env = Environment.capture(0, 0)
env = Environment.capture(1, 0)
env = Environment.capture(0, 1)
Environment.capture(env)
with pytest.raises(TypeError):
Environment.capture("blah")
with pytest.raises(ValueError):
Environment.capture(100)
def test_call_unrecognized_type():
f = lambda x: x
env = Environment.capture()
with pytest.raises(ValueError):
call("f(x)").set_type({"x": 1}, env)
with pytest.raises(ValueError):
call("x").set_type({"x": [1, 2]}, env)
with pytest.raises(ValueError):
call("x").set_type({"x": set([5, 6])}, env)
def test_call_new_data_numeric_stateful_transform():
# The center() transformation remembers the value of the mean
# of the first dataset passed, which is 10.
env = Environment.capture(0)
data = pd.DataFrame({"x": [10, 10, 10]})
call_term = Call(LazyCall("center", [LazyVariable("x")], {}))
call_term.set_type(data, env)
call_term.set_data()
assert (call_term.value == [0, 0, 0]).all()
data = pd.DataFrame({"x": [1, 2, 3]})
assert (call_term.eval_new_data(data) == [-9.0, -8.0, -7.0]).all()
def testevalenv_namespace():
a = 1
b = "hello friend"
c = [1, 2, 3]
d = {"a": 1, "b": 2}
env = Environment.capture()
assert env.namespace["a"] == a
assert env.namespace["b"] == b
assert env.namespace["c"] == c
assert env.namespace["d"] == d
def test_get_function_from_module():
import numpy as np
def function(x):
return x
env = Environment.capture()
NAMES = ["np.random.normal", "np.exp", "function"]
FUNS = [np.random.normal, np.exp, function]
for name, fun in zip(NAMES, FUNS):
f = get_function_from_module(name, env)
assert f == fun
def test_evalenv_with_outer_namespace():
a = 1
b = "hello friend"
env = Environment.capture()
assert env.namespace["a"] == a
assert env.namespace["b"] == b
assert "c" not in env.namespace
assert "d" not in env.namespace
env = env.with_outer_namespace({"c": [1, 2, 3], "d": {"a": 1, "b": 2}})
assert env.namespace["c"] == [1, 2, 3]
assert env.namespace["d"] == {"a": 1, "b": 2}
def test_call_new_data_categoric_stateful_transform():
env = Environment.capture(0)
data = pd.DataFrame({"x": [1, 2, 3]})
# Full rank encoding
call_term = Call(LazyCall("C", [LazyVariable("x")], {}))
call_term.set_type(data, env)
call_term.set_data(spans_intercept=True)
assert (np.array([[1, 0, 0], [0, 1, 0], [0, 0, 1]]) == call_term.value).all()
data = pd.DataFrame({"x": [2, 3]})
assert (call_term.eval_new_data(data) == np.array([[0, 1, 0], [0, 0, 1]])).all()
with pytest.raises(
ValueError,
match=re.escape("The levels 4 in 'C(x)' are not present in the original data set"),
):
data = pd.DataFrame({"x": [2, 3, 4]})
call_term.eval_new_data(data)
# The same with reduced encoding
data = pd.DataFrame({"x": [1, 2, 3]})
call_term = Call(LazyCall("C", [LazyVariable("x")], {}))
call_term.set_type(data, env)
call_term.set_data()
assert (np.array([[0, 0], [1, 0], [0, 1]]) == call_term.value).all()
data = pd.DataFrame({"x": [1, 3]})
assert (call_term.eval_new_data(data) == np.array([[0, 0], [0, 1]])).all()
# It remembers it saw "A", "B", and "C", but not "D".
# So when you pass a new level, it raises a ValueError.
with pytest.raises(
ValueError,
match=re.escape("The levels 4 in 'C(x)' are not present in the original data set"),
):
data = pd.DataFrame({"x": [2, 3, 4]})
call_term.eval_new_data(data)