def test_wines(self):
df = load_wines_dataset()
X = df.drop(['quality', 'color'], axis=1)
X = X[['alcohol', 'volatile_acidity', 'density']]
y = df['quality']
color = df['color']
X_train, X_test, y_train, y_test, color_train, color_test = train_test_split(
X, y, color)
model = SkBaseLearnerCategory("color",
LogisticRegression(solver="liblinear"))
new_x_train = pandas.concat([X_train, color_train], axis=1)
model.fit(new_x_train, y_train)
new_x_test = pandas.concat([X_test, color_test], axis=1)
acc1 = accuracy_score(y_test, model.predict(new_x_test))
try:
self.assertEqualDataFrame(model.models['red'].coef_,
model.models['white'].coef_)
ok = False
except AssertionError as e:
ok = True
self.assertTrue(ok)
clr = LogisticRegression(solver="liblinear")
clr.fit(X_train, y_train)
acc2 = accuracy_score(y_test, clr.predict(X_test))
self.assertGreater(acc1, 0.45)
self.assertGreater(acc2, 0.45)
self.assertGreater(acc1, acc2 * 0.99)
def test_fit_cat_array(self):
df = pandas.DataFrame(
dict(y=[0, 1, 0, 1, 0, 1, 0, 1],
X1=[0.5, 0.6, 0.52, 0.62, 0.5, 0.6, 0.51, 0.61],
X2=[0.5, 0.6, 0.7, 0.5, 1.5, 1.6, 1.7, 1.8],
cat=[9, 10, 9, 10, 9, 10, 9, 10]))
X = df.drop('y', axis=1)
y = df['y']
model = SkBaseLearnerCategory('cat', DecisionTreeClassifier())
model.fit(X, y)
pred = model.predict(X)
self.assertGreater(len(pred), 0)
self.assertEqualArray(y, pred)
def test_fit_predict(self):
df = pandas.DataFrame(
dict(y=[0, 1, 0, 1, 0, 1, 0, 1],
X1=[0.5, 0.6, 0.52, 0.62, 0.5, 0.6, 0.51, 0.61],
X2=[0.5, 0.6, 0.7, 0.5, 1.5, 1.6, 1.7, 1.8],
cat=[
'red', 'red', 'blue', 'blue', 'red', 'red', 'red', 'blue'
]))
X = df.drop('y', axis=1)
y = df['y']
model = SkBaseLearnerCategory('cat', DecisionTreeClassifier())
model.fit(X, y)
pred = model.predict(X)
self.assertGreater(len(pred), 0)
self.assertEqualArray(y, pred)
pred = model.score(X, y)
self.assertGreater(pred, 0)
pred = model.predict_proba(X)
self.assertGreater(len(pred), 0)
self.assertEqualArray(y, pred[:, 1])
self.assertRaise(lambda: model.decision_function(X),
NotImplementedError)
def test_pickle(self):
df = pandas.DataFrame(
dict(y=[0, 1, 0, 1, 0, 1, 0, 1],
X1=[0.5, 0.6, 0.52, 0.62, 0.5, 0.6, 0.51, 0.61],
X2=[0.5, 0.6, 0.7, 0.5, 1.5, 1.6, 1.7, 1.8],
cat=[
'red', 'red', 'blue', 'blue', 'red', 'red', 'red', 'blue'
]))
X = df.drop('y', axis=1)
y = df['y']
model = SkBaseLearnerCategory('cat', DecisionTreeClassifier())
model.fit(X, y)
pred = model.predict(X)
st = BytesIO()
pickle.dump(model, st)
st = BytesIO(st.getvalue())
rec = pickle.load(st)
pred2 = rec.predict(X)
self.assertEqualArray(pred, pred2)
def test_grid(self):
df = pandas.DataFrame(
dict(y=[0, 1, 0, 1, 0, 1, 0, 1],
X1=[0.5, 0.6, 0.52, 0.62, 0.5, 0.6, 0.51, 0.61],
X2=[0.5, 0.6, 0.7, 0.5, 1.5, 1.6, 1.7, 1.8],
cat=[
'red', 'red', 'blue', 'blue', 'red', 'red', 'red', 'blue'
]))
X = df.drop('y', axis=1)
y = df['y']
model = SkBaseLearnerCategory('cat', DecisionTreeClassifier())
res = model.get_params(True)
del res['model']
for c in ['model__ccp_alpha', 'model__presort']:
if c in res:
del res[c]
self.assertEqual(
res, {
'colnameind': 'cat',
'model__class_weight': None,
'model__criterion': 'gini',
'model__max_depth': None,
'model__max_features': None,
'model__max_leaf_nodes': None,
'model__min_impurity_decrease': 0.0,
'model__min_impurity_split': None,
'model__min_samples_leaf': 1,
'model__min_samples_split': 2,
'model__min_weight_fraction_leaf': 0.0,
'model__random_state': None,
'model__splitter': 'best'
})
parameters = {'model__max_depth': [2, 3]}
clf = GridSearchCV(model, parameters, cv=3)
clf.fit(X, y)
pred = clf.predict(X)
self.assertEqualArray(y, pred)