def test_info_gain_one_target_one_feature(self, idadf):
if len(idadf.columns) > 1:
result = info_gain(idadf,
target=idadf.columns[0],
features=[idadf.columns[1]])
assert (isinstance(result, float))
result2 = info_gain(idadf,
target=idadf.columns[1],
features=[idadf.columns[0]])
assert (round(result, 3) == round(result2, 3)) # symmetry
def test_info_gain_default(self, idadf):
if len(idadf.columns) > 1:
result = info_gain(idadf, features = idadf.columns)
assert(isinstance(result, pandas.core.frame.DataFrame))
assert(len(result.columns) == len(idadf.columns))
assert(len(result.index) == len(idadf.columns))
result2 = info_gain(idadf)
assert(all(result == result2))
result = result.fillna(0) # np.nan values are not equal when compared
assert(all(result == result.T)) # symmetry
def test_info_gain_default(self, idadf):
if len(idadf.columns) > 1:
result = info_gain(idadf, features=idadf.columns)
assert (isinstance(result, pandas.core.frame.DataFrame))
assert (len(result.columns) == len(idadf.columns))
assert (len(result.index) == len(idadf.columns))
result2 = info_gain(idadf)
assert (all(result == result2))
result = result.fillna(
0) # np.nan values are not equal when compared
assert (all(result == result.T)) # symmetry
def test_info_gain_valueError(self, idadf):
if len(idadf.columns) > 0:
with pytest.raises(
ValueError
): # Cannot compute correlation coefficients of only one column (...), need at least 2
info_gain(idadf, features=idadf.columns[0])
with pytest.raises(
ValueError
): # The correlation value of two same columns is always maximal
info_gain(idadf,
target=idadf.columns[0],
features=idadf.columns[0])
def test_info_gain_multiple_target(self, idadf):
if len(idadf.columns) > 1:
result = info_gain(idadf,
target=[idadf.columns[0], idadf.columns[1]])
assert (isinstance(result, pandas.core.frame.DataFrame))
assert (len(result.columns) == 2)
assert (len(result.index) == len(idadf.columns))
def test_info_gain_one_target_one_feature(self, idadf):
if len(idadf.columns) > 1:
result = info_gain(idadf, target = idadf.columns[0], features=[idadf.columns[1]])
assert(isinstance(result, float))
result2 = info_gain(idadf, target = idadf.columns[1], features=[idadf.columns[0]])
assert(round(result,3) == round(result2,3)) # symmetry
def test_info_gain_multiple_target(self, idadf):
if len(idadf.columns) > 1:
result = info_gain(idadf, target = [idadf.columns[0],idadf.columns[1]])
assert(isinstance(result, pandas.core.frame.DataFrame))
assert(len(result.columns) == 2)
assert(len(result.index) == len(idadf.columns))
def test_info_gain_one_target(self, idadf):
if len(idadf.columns) > 1:
result = info_gain(idadf, target = idadf.columns[0])
assert(isinstance(result, pandas.core.series.Series))
assert(len(result) == (len(idadf.columns)-1))
def test_info_gain_valueError(self, idadf):
if len(idadf.columns) > 0:
with pytest.raises(ValueError): # Cannot compute correlation coefficients of only one column (...), need at least 2
info_gain(idadf, features = idadf.columns[0])
with pytest.raises(ValueError): # The correlation value of two same columns is always maximal
info_gain(idadf, target= idadf.columns[0], features = idadf.columns[0])
def apply_Model(X, Y):
# X_train=X_train.iloc[:,0:1139]
idadb = IdaDataBase("DASHDB")
idadf = IdaDataFrame(idadb)
IG = info_gain(idadb, target=Y, features=X)
def test_info_gain_one_target(self, idadf):
if len(idadf.columns) > 1:
result = info_gain(idadf, target=idadf.columns[0])
assert (isinstance(result, pandas.core.series.Series))
assert (len(result) == (len(idadf.columns) - 1))
