def test_keepothers(self):
sel = notation((Select('ID')))
r = sel.fit_transform(self.df)
self.assertEqual(len(r.columns.tolist()), 1)
sel = notation((Select('ID'), KeepOthers()))
r = sel.fit_transform(self.df)
all_columns = self.df.columns.tolist()
self.assertSetEqual(set(all_columns), set(r.columns.tolist()))
def test_notation(self):
pl1 = notation([Impute(0)])
self.assertTrue(isinstance(pl1, Pipeline))
pl2 = notation([[[Impute(0)]]])
self.assertTrue(isinstance(pl2, Pipeline))
self.assertTrue(isinstance(pl2[0], Pipeline))
self.assertTrue(isinstance(pl2[0][0], Pipeline))
pl3 = notation(([Impute(0)], Select('age')))
self.assertTrue(isinstance(pl3, PipelineUnion))
self.assertTrue(isinstance(pl3[0], Pipeline))
def test_drop_columns(self):
drop_cols = ['sex', 'ID']
drop_cols = notation(([Select(drop_cols), Drop()], KeepOthers()))
r = drop_cols.fit_transform(self.df)
r_cols = r.columns.tolist()
for col in drop_cols:
self.assertTrue(col not in r_cols)
def test_impute_zero(self):
series = self.df['age']
msk_na = pd.isnull(series)
impute = notation([Select('age'), Impute(0)])
r = impute.fit_transform(self.df)
n_notzero = (r.loc[msk_na, 'age'] != 0).sum()
self.assertTrue(n_notzero == 0)
def test_Scale(self):
col = 'height(cm)'
min_ = 0
max_ = 1
self.assertGreater(self.df[col].max(), max_)
ppl = notation([Select(col), Scale(min_, max_)])
r = ppl.fit_transform(self.df)
self.assertEqual(r[col].max(), max_)
self.assertEqual(r[col].min(), min_)