def test_regret_equal():
"""Tests instances where trials share best objectives"""
data = pd.DataFrame(
data={
"id": ["a", "b", "c"],
"objective": [8, 9, 8],
}
)
expected_best = [8, 8, 8]
expected_ids = ["a", "a", "a"]
result = regret(data)
assert all(result["best"].values == expected_best)
assert all(result["best_id"].values == expected_ids)
data = pd.DataFrame(
data={
"id": ["a", "b", "c"],
"objective": [8, 8, 8],
}
)
expected_best = [8, 8, 8]
expected_ids = ["a", "a", "a"]
result = regret(data)
assert all(result["best"].values == expected_best)
assert all(result["best_id"].values == expected_ids)
def test_regret_sequential():
"""Tests that cumulative best objectives are linked to their respective ids"""
data = pd.DataFrame(
data={
"id": ["a", "b", "c", "d"],
"objective": [10, 12, 8, 9],
}
)
expected_best = [10, 10, 8, 8]
expected_ids = ["a", "a", "c", "c"]
result = regret(data)
assert all(result["best"].values == expected_best)
assert all(result["best_id"].values == expected_ids)
data = pd.DataFrame(
data={
"id": ["a", "b", "c"],
"objective": [10, 9, 8],
}
)
expected_best = [10, 9, 8]
expected_ids = ["a", "b", "c"]
result = regret(data)
assert all(result["best"].values == expected_best)
assert all(result["best_id"].values == expected_ids)
data = pd.DataFrame(
data={
"id": ["a", "b", "c"],
"objective": [8, 9, 10],
}
)
expected_best = [8, 8, 8]
expected_ids = ["a", "a", "a"]
result = regret(data)
assert all(result["best"].values == expected_best)
assert all(result["best_id"].values == expected_ids)
data = pd.DataFrame(
data={
"id": ["a"],
"objective": [10],
}
)
result = regret(data)
assert all(result["best"].values == [10])
assert all(result["best_id"].values == ["a"])
def test_length_names():
"""Tests the length of `names` parameter is two"""
data = pd.DataFrame(data={"id": ["a", "b"], "objective": [0.1, 0.2]})
with pytest.raises(ValueError) as exception:
regret(data, names=())
assert "`names` requires a tuple with 2 elements. 0 provided." in str(
exception.value
)
with pytest.raises(ValueError) as exception:
regret(data, names=("a", "b", "c"))
assert "`names` requires a tuple with 2 elements. 3 provided." in str(
exception.value
)
def test_accept_empty():
"""Tests an empty dataframe is returned if you give an empty dataframe"""
empty_frame = pd.DataFrame()
result = regret(empty_frame)
assert result.empty
assert result.equals(empty_frame)
empty_frame = pd.DataFrame(columns=["id", "objective"])
result = regret(empty_frame)
assert result.empty
assert "best" not in result.columns
assert "best_id" not in result.columns
assert result.equals(empty_frame)
def test_accept_dict():
"""Tests dictionary parameter is supported"""
data = {"id": ["a", "b", "c", "d"], "objective": [0.1, 0.2, 0.3, 0.5]}
result = regret(data)
assert type(result) is pd.DataFrame
def test_input_column_names():
"""Tests custom column names"""
data = pd.DataFrame(
data={"id": ["a", "b", "c", "d"], "objective": [0.1, 0.2, 0.3, 0.5]}
)
result = regret(data, names=("something", "else"))
assert "something" in result.columns
assert "else" in result.columns
def test_default_column_names():
"""Tests default column names"""
data = pd.DataFrame(
data={"id": ["a", "b", "c", "d"], "objective": [0.1, 0.2, 0.3, 0.5]}
)
result = regret(data)
assert "best" in result.columns
assert "best_id" in result.columns
def test_parameter_not_modified():
"""Tests the original dataframe is not modified"""
data = pd.DataFrame(
data={"id": ["a", "b", "c", "d"], "objective": [0.1, 0.2, 0.3, 0.5]}
)
result = regret(data)
assert len(data.columns) == 2
assert len(result.columns) == 4