def test_data_splitting_test_ratio(df_10_rows):
train, valid, test = data_splitting.split(df_10_rows,
train_ratio=0.7,
valid_ratio=0.2,
test_ratio=0.1,
seed=0)
assert len(test) == 1
def test_data_splitting_valid_vs_test(df_10_rows):
train, valid, test = data_splitting.split(df_10_rows,
train_ratio=0.7,
valid_ratio=0.2,
test_ratio=0.1,
seed=0)
df_check = valid.merge(
test,
how="inner",
right_on=["col_1", "col_2"],
left_on=["col_1", "col_2"],
sort=False,
)
assert df_check.empty