def test_both_are_pipelines():
group_1 = compose.Select("a",
"b") | preprocessing.OneHotEncoder(sparse=True)
group_2 = compose.Select("x",
"y") | preprocessing.OneHotEncoder(sparse=True)
product = group_1 + group_2 + group_1 * group_2
assert product.transform_one(dict(a=1, b=2, x=4, y=4, z=5)) == {
"b_2*x_4": 1,
"b_2*y_4": 1,
"a_1*x_4": 1,
"a_1*y_4": 1,
"x_4": 1,
"y_4": 1,
"b_2": 1,
"a_1": 1,
}
def test_left_is_pipeline():
group_1 = compose.Select("a", "b")
group_2 = compose.Select("x",
"y") | preprocessing.OneHotEncoder(sparse=True)
product = group_1 + group_2 + group_1 * group_2
assert product.transform_one(dict(a=1, b=2, x=4, y=4, z=5)) == {
"a*y_4": 1,
"a*x_4": 1,
"b*y_4": 2,
"b*x_4": 2,
"y_4": 1,
"x_4": 1,
"a": 1,
"b": 2,
}
def __iter__(self):
oh = (compose.SelectType(str) | preprocessing.OneHotEncoder()) + compose.SelectType(
int
)
for x, y in datasets.SolarFlare().take(200):
yield oh.transform_one(x), y