def test_fail_if_missing_pid(minimal_input_data):
data = minimal_input_data.drop("p_id", axis=1).copy()
with pytest.raises(
ValueError,
match="The input data must contain the column p_id",
):
compute_taxes_and_transfers(data, {}, functions=[], targets=[])
def test_fail_if_non_unique_pid(minimal_input_data):
data = minimal_input_data.copy()
data["p_id"] = 1
with pytest.raises(
ValueError,
match="The following p_ids are non-unique",
):
compute_taxes_and_transfers(data, {}, functions=[], targets=[])
def test_function_without_data_dependency_is_not_mistaken_for_data():
n_individuals = 5
df = pd.DataFrame(index=np.arange(n_individuals))
def a():
return pd.Series(range(n_individuals))
def b(a):
return a
compute_taxes_and_transfers(df, {}, functions=[a, b], targets="b")
def test_function_without_data_dependency_is_not_mistaken_for_data(
minimal_input_data):
def a():
return pd.Series(range(minimal_input_data.shape[0]))
def b(a):
return a
compute_taxes_and_transfers(minimal_input_data, {},
functions=[a, b],
targets="b")
def test_missing_root_nodes_raises_error(minimal_input_data):
def b(a):
return a
def c(b):
return b
with pytest.raises(
ValueError,
match="The following data columns are missing",
):
compute_taxes_and_transfers(minimal_input_data, {},
functions=[b, c],
targets="c")
def test_missing_root_nodes_raises_error():
n_individuals = 5
df = pd.DataFrame(index=np.arange(n_individuals))
def b(a):
return a
def c(b):
return b
with pytest.raises(
ValueError,
match="The following data columns are missing",
):
compute_taxes_and_transfers(df, {}, functions=[b, c], targets="c")
def test_pension(input_data, year):
column = "rente_anspr_m"
year_data = input_data[input_data["jahr"] == year]
df = year_data[INPUT_COLS].copy()
policy_params, policy_functions = set_up_policy_environment(
date=f"{year}-07-01")
calc_result = compute_taxes_and_transfers(
data=df,
params=policy_params,
functions=policy_functions,
targets=column,
)
assert_series_equal(calc_result[column].round(2), year_data[column])
def test_update_earning_points(input_data, year):
year_data = input_data[input_data["jahr"] == year]
df = year_data[INPUT_COLS].copy()
policy_params, policy_functions = set_up_policy_environment(
date=f"{year}-07-01")
calc_result = compute_taxes_and_transfers(
data=df,
params=policy_params,
functions=policy_functions,
targets="entgeltpunkte_update",
)
assert_series_equal(calc_result["entgeltpunkte_update"],
year_data["EP_end"],
check_names=False)
def test_synthetic():
"""
Test creation of synthetic data
"""
# run with defaults
df = create_synthetic_data()
# rent must be positive
assert df["kaltmiete_m_hh"].min() > 0
# heating cost must be positive
assert df["heizkosten_m_hh"].min() > 0
# no NaN values
assert df.notna().all().all()
# correct dimensions for every household type
assert len(df[df["hh_typ"] == "couple_0_children"] == 2)
assert len(df[df["hh_typ"] == "single_2_children"] == 3)
assert len(df[df["hh_typ"] == "couple_2_children"] == 4)
# unique personal id?
assert df["p_id"].is_unique
doppelverdiener = create_synthetic_data(hh_typen=["couple"],
n_children=[0],
double_earner=True,
bruttolohn_m=2000)
assert (doppelverdiener["bruttolohn_m"] > 0).all()
# test heterogeneity
incrange = create_synthetic_data(
hh_typen=["couple"],
n_children=0,
heterogeneous_vars={
"bruttolohn_m": list(np.arange(0, 6000, 1000)),
"vermögen_hh": [10_000, 500_000, 1_000_000],
},
)
# is household id unique?
assert (incrange.groupby("hh_id").size() == 2).all()
assert incrange.notna().all().all()
# finally, run through gettsim
policy_params, policy_functions = set_up_policy_environment(2020)
results = compute_taxes_and_transfers(df, policy_params, policy_functions)
assert len(results) == len(df)