def test_eda_raise_error_not_dataframe():
test_input = [1, 2, 3, 4, 5]
message = "Input data must be an instance of DataFrame"
with pytest.raises(TypeError, match=message):
eda.eda(test_input, "target")
def test_nb_classes():
test_data = gen_test_data()
eda_res = eda.eda(test_data, "target")
message = "Wrong output for the number of class lables"
assert eda_res["nb_class"] == 3, message
def test_pairplot():
test_data = gen_test_data()
eda_res = eda.eda(test_data, "target")
p = eda_res["pairplot"]
message = 'Wrong output for altair plot'
assert isinstance(p, alt.vegalite.v4.api.RepeatChart), message
def test_len_cat_features():
test_data = gen_test_data()
eda_res = eda.eda(test_data, "target")
message = "Wrong output for the length of categorical features"
assert eda_res["nb_cat_features"] == 1, message
def test_cat_feature():
test_data = gen_test_data()
eda_res = eda.eda(test_data, "target")
message = "Wrong output for categorical feature names"
assert "cat1" in eda_res["cat_features_name"], message
def test_len_num_feature():
test_data = gen_test_data()
eda_res = eda.eda(test_data, "target")
message = "Wrong output for the length of numerical features"
assert eda_res["nb_num_features"] == 4, message
def test_num_feature():
test_data = gen_test_data()
eda_res = eda.eda(test_data, "target")
message = "Wrong output for numerical feature names"
assert "num2" in eda_res["num_features_name"], message
def test_attributes():
test_data = gen_test_data()
eda_res = eda.eda(test_data, "target")
assert 'cat_features_name' in eda_res
assert 'num_features_name' in eda_res
assert 'pairplot' in eda_res
