def test_parser():
train_df = pd.DataFrame({
"A": [i for i in range(100)],
"B": ["hello" if i % 2 == 0 else "bye" for i in range(100)],
})
params = {"train_df": train_df, "target_label": "C"}
encoder = Encoder()
encoder.encode(params=params)
assert "B" in params["cat_cols"]
def test_categorical_encoding():
train_csv = pd.read_csv("datasets/encoding/testnew.csv")
params = {
"train_df": train_csv,
"target_label": "Price",
"cat_cols": ["Profession"],
"one_hot": False,
}
encoder = Encoder()
encoder.encode(params=params)
assert "ProfessionEncoded" in params["train_df"].columns
assert params["train_df"]["ProfessionEncoded"][0] == 0
def test_ignore_cat_col():
train_csv = pd.read_csv("datasets/encoding/testnew.csv")
params = {
"train_df": train_csv,
"target_label": "Price",
"cat_cols": ["Profession"],
"ord_dict": ord_dict,
"one_hot": True,
}
encoder = Encoder()
encoder.encode(params=params)
assert "Profession_HOD" not in params["train_df"].columns
def test_one_hot_encoding():
train_csv = pd.read_csv("datasets/encoding/testnew.csv")
params = {
"train_df": train_csv,
"target_label": "Price",
"cat_cols": ["Test", "Labels"],
"ord_dict": ord_dict,
"one_hot": True,
}
encoder = Encoder()
encoder.encode(params=params)
assert "Test_Tata" in params["train_df"].columns
assert params["train_df"]["Test_Tata"][1] == 1
def test_empty_weight_mapping():
train_csv = pd.read_csv("datasets/encoding/testnew.csv")
train_csv.drop(["Price"], axis=1, inplace=True)
ord_dict1 = ord_dict.copy()
ord_dict1["Size"] = None
params = {
"train_df": train_csv,
"target_label": "Price",
"ord_dict": ord_dict1,
}
with pytest.raises(ValueError):
encoder = Encoder()
encoder.encode(params=params)
def test_mapping():
train_csv = pd.read_csv("datasets/encoding/testnew.csv")
train_csv.drop(["Price"], axis=1, inplace=True)
params = {
"train_df": train_csv,
"target_label": "Price",
"ord_dict": ord_dict,
}
encoder = Encoder()
encoder.encode(params=params)
assert params["train_df"]["ProfessionEncoded"].nunique() == 3
assert params["train_df"]["ProfessionEncoded"][2] == 3
assert Counter(params["ord_dict"]["Profession"].values()) == Counter(
params["train_df"]["ProfessionEncoded"].unique())
def __init__(
self,
train_df_path=None,
test_df_path=None,
steps=None,
config_file=None,
params=None,
custom_reader=None,
):
steps = [
Parser().parse_dataset,
NullValuesHandler().execute,
Encoder().encode,
HandleOutlier().handle_outliers,
Scaler().execute,
SelectKBest().fit_transform,
Split().train_test_split,
]
super().__init__(
train_df_path=train_df_path,
test_df_path=test_df_path,
steps=steps,
config_file=config_file,
params=params,
custom_reader=custom_reader,
)
def test_target_label_warning():
train_csv = pd.read_csv("datasets/encoding/testnew.csv")
params = {"train_df": train_csv, "ord_dict": ord_dict}
with pytest.warns(UserWarning):
encoder = Encoder()
encoder.encode(params=params)
def test_empty_df():
params = {"target_label": "Price", "ord_dict": ord_dict}
with pytest.raises(ValueError):
encoder = Encoder()
encoder.encode(params=params)