def test_init_df_with_extra_col(test_pd_df):
tdf = test_pd_df.copy()
extra_col = "climate model"
extra_value = "scm_model"
tdf[extra_col] = extra_value
df = IamDataFrame(tdf)
assert df.extra_cols == [extra_col]
pd.testing.assert_frame_equal(df.timeseries().reset_index(),
tdf, check_like=True)
def test_relationship_negative_specific(self, unequal_df,
test_downscale_df, match_sign,
caplog):
# Test that match_sign results in the correct multipliers when negative values
# are added to the positive ones.
follow = "Emissions|HFC|C5F12"
lead = ["Emissions|HFC|C2F6"]
equal_df = unequal_df.filter(model="model_a")
invert_sign = equal_df.filter(variable=lead)
invert_sign.data["value"] = -1 * invert_sign.data["value"]
invert_sign.append(equal_df.filter(variable=follow), inplace=True)
invert_sign.data["model"] = "negative_model"
equal_df.append(invert_sign, inplace=True)
equal_df = IamDataFrame(equal_df.data)
tcruncher = self.tclass(equal_df)
# Invert the sign of the infillee too, as we haven't tested the formula for
# negative values
test_downscale_df.data["value"] = -1 * test_downscale_df.data["value"]
test_downscale_df = self._adjust_time_style_to_match(
test_downscale_df, equal_df).filter(year=[2010, 2015])
filler = tcruncher.derive_relationship(variable_follower=follow,
variable_leaders=lead,
same_sign=match_sign)
res = filler(test_downscale_df)
# if we have match sign on, this is identical to the above except for -ve.
if match_sign:
lead_iamdf = test_downscale_df.filter(
variable="Emissions|HFC|C2F6")
exp = lead_iamdf.timeseries()
# The follower values are 1 and 9 (average 5), the leader values
# are all -1, hence we expect the input * -5 as output.
exp[exp.columns[0]] = exp[exp.columns[0]] * -5
exp[exp.columns[1]] = exp[exp.columns[1]] * -1
exp = exp.reset_index()
exp["variable"] = "Emissions|HFC|C5F12"
exp["unit"] = "kt C5F12/yr"
exp = IamDataFrame(exp)
# Test that our constructed value is the same as the result
pd.testing.assert_frame_equal(res.timeseries(),
exp.timeseries(),
check_like=True)
# comes back on input timepoints
np.testing.assert_array_equal(
res.timeseries().columns.values.squeeze(),
test_downscale_df.timeseries().columns.values.squeeze(),
)
else:
# If we combine all signs together, there is no net lead but a net follow,
# so we have have a multiplier of infinity.
assert all(res.data["value"] == -np.inf)
def test_init_df_with_extra_col(test_pd_df):
tdf = test_pd_df.copy()
extra_col = "climate model"
extra_value = "scm_model"
tdf[extra_col] = extra_value
df = IamDataFrame(tdf)
# check that timeseries data is as expected
obs = df.timeseries().reset_index()
exp = tdf[obs.columns] # get the columns into the right order
pd.testing.assert_frame_equal(obs, exp)
def test_relationship_usage_interpolation(self, test_db, test_downscale_df,
interpolate):
tcruncher = self.tclass(test_db)
filler = tcruncher.derive_relationship("Emissions|HFC|C5F12",
["Emissions|HFC|C2F6"])
test_downscale_df = self._adjust_time_style_to_match(
test_downscale_df.filter(year=2015, keep=False), test_db)
required_timepoint = test_db.filter(
year=2015).data[test_db.time_col].iloc[0]
if not interpolate:
if isinstance(required_timepoint, pd.Timestamp):
required_timepoint = required_timepoint.to_pydatetime()
error_msg = re.escape(
"Required downscaling timepoint ({}) is not in the data for the "
"lead gas (Emissions|HFC|C2F6)".format(required_timepoint))
with pytest.raises(ValueError, match=error_msg):
filler(test_downscale_df, interpolate=interpolate)
return
res = filler(test_downscale_df, interpolate=interpolate)
lead_iamdf = test_downscale_df.filter(variable="Emissions|HFC|C2F6",
region="World",
unit="kt C2F6/yr")
exp = lead_iamdf.timeseries()
# will have to make this more intelligent for time handling
lead_df = lead_iamdf.timeseries()
lead_df[required_timepoint] = np.nan
lead_df = lead_df.reindex(sorted(lead_df.columns), axis=1)
lead_df = lead_df.interpolate(method="index", axis=1)
lead_val_2015 = lead_df[required_timepoint]
exp = (exp.T * 3.14 / lead_val_2015).T.reset_index()
exp["variable"] = "Emissions|HFC|C5F12"
exp["unit"] = "kt C5F12/yr"
exp = IamDataFrame(exp)
pd.testing.assert_frame_equal(res.timeseries(),
exp.timeseries(),
check_like=True)
# comes back on input timepoints
np.testing.assert_array_equal(
res.timeseries().columns.values.squeeze(),
test_downscale_df.timeseries().columns.values.squeeze(),
)
def test_relationship_usage(self, test_downscale_df, add_col):
units = "new units"
tcruncher = self.tclass()
test_downscale_df = test_downscale_df.filter(year=[2010, 2015])
if add_col:
# what should happen if there's more than one value in the `add_col`?
add_col_val = "blah"
test_downscale_df[add_col] = add_col_val
test_downscale_df = IamDataFrame(test_downscale_df.data)
assert test_downscale_df.extra_cols[0] == add_col
lead = ["Emissions|HFC|C2F6"]
follow = "Emissions|HFC|C5F12"
filler = tcruncher.derive_relationship(follow,
lead,
ratio=2,
units=units)
res = filler(test_downscale_df)
exp = test_downscale_df.filter(variable=lead)
exp.data["variable"] = follow
exp.data["value"] = exp.data["value"] * 2
exp.data["unit"] = units
pd.testing.assert_frame_equal(res.timeseries(),
exp.timeseries(),
check_like=True)
# comes back on input timepoints
np.testing.assert_array_equal(
res.timeseries().columns.values.squeeze(),
test_downscale_df.timeseries().columns.values.squeeze(),
)
# Test we can append the results correctly
append_df = test_downscale_df.append(res)
assert append_df.filter(variable=follow).equals(res)
if add_col:
assert all(append_df.filter(variable=lead)[add_col] == add_col_val)
def test_init_df_from_timeseries(test_df):
df = IamDataFrame(test_df.timeseries())
pd.testing.assert_frame_equal(df.timeseries(), test_df.timeseries())
def test_init_df_with_index(test_pd_df):
df = IamDataFrame(test_pd_df.set_index(META_IDX))
pd.testing.assert_frame_equal(df.timeseries().reset_index(), test_pd_df)
