def test_can_add_hoc_lv_paths_correctly():
structure = c.Structure()
structure.add_path(["MANDRILL", "BONOBO"], ["APE"])
structure.add_path(["APE"], ["GOAT"])
initial_path = structure.path()
config = c.Config(initial_path)
config.add_higher_order("APE", Mode.A, ["CHEDDAR", "GOUDA"])
estimator = Estimator(config)
structure = c.Structure(initial_path)
structure.add_path(["MANDRILL", "BONOBO"], ["CHEDDAR"])
structure.add_path(["MANDRILL", "BONOBO"], ["GOUDA"])
structure.add_path(["GOUDA", "CHEDDAR"], ["GOAT"])
expected = structure.path().drop("APE").drop("APE", axis=1)
actual = estimator.hoc_path_first_stage(config)
pt.assert_frame_equal(expected, actual)
def test_paths():
mobi = pd.read_csv("file:tests/data/mobi.csv", index_col=0)
structure = c.Structure()
structure.add_path(["Expectation", "Quality"], ["Loyalty"])
structure.add_path(["Image"], ["Expectation"])
structure.add_path(["Complaints"], ["Loyalty"])
config = c.Config(structure.path(), default_scale=Scale.NUM)
config.add_lv_with_columns_named("Expectation", Mode.A, mobi, "CUEX")
config.add_lv_with_columns_named("Quality", Mode.B, mobi, "PERQ")
config.add_lv_with_columns_named("Loyalty", Mode.A, mobi, "CUSL")
config.add_lv_with_columns_named("Image", Mode.A, mobi, "IMAG")
config.add_lv_with_columns_named("Complaints", Mode.A, mobi, "CUSCO")
mobi_pls = Plspm(mobi, config, Scheme.PATH, 100, 0.00000001)
expected_outer_model = pd.read_csv(
"file:tests/data/seminr-mobi-basic-outer-model.csv", index_col=0)
actual_outer_model = mobi_pls.outer_model().drop(
["communality", "redundancy"], axis=1)
npt.assert_allclose(expected_outer_model.sort_index(),
actual_outer_model.sort_index(),
rtol=1e-5)
expected_paths = pd.read_csv("file:tests/data/seminr-mobi-basic-paths.csv",
index_col=0)
actual_paths = mobi_pls.path_coefficients().transpose()
npt.assert_allclose(expected_paths.sort_index().sort_index(axis=1),
actual_paths.sort_index().sort_index(axis=1),
rtol=1e-6)
def test_cannot_add_mvs_twice():
structure = c.Structure()
structure.add_path(source=["BONOBO"], target=["APE"])
config = c.Config(structure.path())
config.add_lv("BONOBO", Mode.A, c.MV("a"), c.MV("b"))
with pytest.raises(ValueError):
config.add_lv("APE", Mode.A, c.MV("a"), c.MV("b"))
def satisfaction_path_matrix():
structure = c.Structure()
structure.add_path(["IMAG"], ["EXPE", "SAT", "LOY"])
structure.add_path(["EXPE"], ["QUAL", "VAL", "SAT"])
structure.add_path(["QUAL"], ["VAL", "SAT"])
structure.add_path(["VAL"], ["SAT"])
structure.add_path(["SAT"], ["LOY"])
return structure.path()
def test_create_path_from_structure():
lvs = ["MANDRILL", "BONOBO", "APE", "GOAT", "CATFISH"]
expected = pd.DataFrame([[0, 0, 0, 0, 0], [0, 0, 0, 0, 0], [1, 1, 0, 0, 0],
[0, 0, 1, 0, 0], [0, 0, 1, 0, 0]],
index=lvs,
columns=lvs)
structure = c.Structure()
structure.add_path(source=["BONOBO", "MANDRILL"], target=["APE"])
structure.add_path(source=["APE"], target=["CATFISH", "GOAT"])
pt.assert_frame_equal(expected, structure.path())
def hoc_path_first_stage(self, config: c.Config) -> pd.DataFrame:
# For first pass, for HOCs we'll create paths from each and for each exogenous LV to the HOC's constituent LVs,
# and from each consituent LV to the endogenous LVs.
path = config.path()
for hoc, lvs in config.hoc().items():
structure = c.Structure(path)
exogenous = path.loc[hoc]
endogenous = path.loc[:, hoc]
# structure.add_path(lvs, [hoc])
for lv in list(exogenous[exogenous == 1].index):
structure.add_path([lv], lvs)
for lv in list(endogenous[endogenous == 1].index):
structure.add_path(lvs, [lv])
path = structure.path().drop(hoc).drop(hoc, axis=1)
return path
def test_hoc_two_stage():
mobi = pd.read_csv("file:tests/data/mobi.csv", index_col=0)
structure = c.Structure()
structure.add_path(["Expectation", "Quality"], ["Satisfaction"])
structure.add_path(["Satisfaction"], ["Complaints", "Loyalty"])
config = c.Config(structure.path(), default_scale=Scale.NUM)
config.add_higher_order("Satisfaction", Mode.A, ["Image", "Value"])
config.add_lv_with_columns_named("Expectation", Mode.A, mobi, "CUEX")
config.add_lv_with_columns_named("Quality", Mode.B, mobi, "PERQ")
config.add_lv_with_columns_named("Loyalty", Mode.A, mobi, "CUSL")
config.add_lv_with_columns_named("Image", Mode.A, mobi, "IMAG")
config.add_lv_with_columns_named("Complaints", Mode.A, mobi, "CUSCO")
config.add_lv_with_columns_named("Value", Mode.A, mobi, "PERV")
mobi_pls = Plspm(mobi, config, Scheme.PATH, 100, 0.00000001)
expected_outer_model = pd.read_csv(
"file:tests/data/seminr-mobi-hoc-ts-outer-model.csv", index_col=0)
actual_outer_model = mobi_pls.outer_model().drop(
["communality", "redundancy"], axis=1)
indices = list(
set(expected_outer_model.index.values.tolist()).intersection(
set(actual_outer_model.index.values.tolist())))
expected_outer_model = expected_outer_model.loc[indices].sort_index(
).sort_index(axis=1)
actual_outer_model = actual_outer_model.loc[indices].sort_index(
).sort_index(axis=1)
npt.assert_allclose(expected_outer_model, actual_outer_model, rtol=1e-4)
expected_paths = pd.read_csv(
"file:tests/data/seminr-mobi-hoc-ts-paths.csv",
index_col=0).transpose()
actual_paths = mobi_pls.path_coefficients()
npt.assert_allclose(expected_paths.sort_index().sort_index(axis=1),
actual_paths.sort_index().sort_index(axis=1),
rtol=1e-6)
def russa_path_matrix():
structure = c.Structure()
structure.add_path(["AGRI", "IND"], ["POLINS"])
return structure.path()