def test_already_index(self):
features = pd.DataFrame({
"cat": [0, 1, 0, 0, 2, np.nan, 1],
"num": [1, 2, 3, 4, 5, 6, 7]
})
encoder = utils.StringIndexer(cols=["cat"])
encoder.fit(features)
pd.testing.assert_series_equal(encoder.transform(features)["cat"],
pd.Series([1, 2, 1, 1, 3, 0, 2]),
check_names=False)
def test_topk(self):
features = pd.DataFrame({
"cat": ["a", "b", "a", "a", "d", "c", "b"],
"num": [1, 2, 3, 4, 5, 6, 7]
})
encoder = utils.StringIndexer(cols=["cat"], topk=2)
encoder.fit(features)
pd.testing.assert_series_equal(encoder.transform(features)["cat"],
pd.Series([1, 2, 1, 1, 3, 3, 2]),
check_names=False)
def test_realdata(self):
bunch = load_boston()
features = pd.DataFrame(bunch.data, columns=bunch.feature_names)
encoder = utils.StringIndexer(cols=["CHAS", "RAD"])
encoder.fit(features)
truth_chas = copy.deepcopy(features["CHAS"])
def chas_map(x):
if x == 0.0:
return 1
elif x == 1.0:
return 2
truth_chas = truth_chas.map(chas_map)
truth_rad = copy.deepcopy(features["RAD"])
def rad_map(x):
if x == 24.0:
return 1
elif x == 5.0:
return 2
elif x == 4.0:
return 3
elif x == 3.0:
return 4
elif x == 6.0:
return 5
elif x == 8.0:
return 6
elif x == 2.0:
return 7
elif x == 1.0:
return 8
elif x == 7.0:
return 9
truth_rad = truth_rad.map(rad_map)
transformed = encoder.transform(features)
pd.testing.assert_series_equal(transformed["CHAS"],
truth_chas,
check_names=False)
pd.testing.assert_series_equal(transformed["RAD"],
truth_rad,
check_names=False)