def test_fit_single_model_with_2_class_proba(self):
layer_model = Layer([LogisticRegression(solver='liblinear')],
proba=True)
X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
y = np.array([1, 1, 0, 0])
result = layer_model.fit(X, y)
assert result.shape == (4, 2)
def test_predict_single_model_with_preprocess(self):
layer_model = Layer([LinearRegression()], [MinMaxScaler()])
X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
y = np.dot(X, np.array([1, 2])) + 3
layer_model.fit(X, y)
result = layer_model.predict(np.array([[3, 5]]))
assert result.shape == (1, 1)
assert np.allclose(result, np.array([[16]]))
def test_fir_single_model_with_preprocess(self):
layer_model = Layer([LinearRegression()], [MinMaxScaler()])
X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
y = np.dot(X, np.array([1, 2])) + 3
# X and y are linearly related, predictions will be almost perfect
result = layer_model.fit(X, y)
assert result.shape == (4, 1)
assert np.allclose(result.flatten(), y)
def test_fit_single_model_with_multi_class_proba(self):
layer_model = Layer(
[LogisticRegression(solver='lbfgs', multi_class='multinomial')],
proba=True)
X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
y = np.array([1, 1, 0, 2])
result = layer_model.fit(X, y)
assert result.shape == (4, 3)
def test_predict_multiple_model_with_2_class_proba(self):
layer_model = Layer([LogisticRegression(solver='liblinear'),
LogisticRegression(solver='liblinear')],
proba=[True,False])
X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
y = np.array([1, 1, 0, 0])
layer_model.fit(X, y)
result = layer_model.predict(np.array([[3, 5], [2, 5]]))
assert result.shape == (2,3)
def test_predict_multiple_model(self):
layer_model = Layer([LinearRegression(), LinearRegression()],
[None, MinMaxScaler()])
X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
y = np.dot(X, np.array([1, 2])) + 3
layer_model.fit(X, y)
result = layer_model.predict(np.array([[3, 5]]))
assert result.shape == (1,2)
assert np.allclose(result, np.array([[16, 16]]))
def test_using_proba_without_predict_proba_method(self):
with pytest.warns(Warning) as record:
layer_model = Layer([LinearRegression()], proba=True)
X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
y = np.dot(X, np.array([1, 2])) + 3
layer_model.fit(X, y)
result = layer_model.predict(np.array([[3, 5], [3, 5]]))
assert result.shape == (2, 1)
assert np.allclose(result, np.array([[16], [16]]))
assert record
def test_fit_multiple_models(self):
layer_model = Layer([LinearRegression(), LinearRegression()],
[None, MinMaxScaler()])
X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
y = np.dot(X, np.array([1, 2])) + 3
# X and y are linearly related, predictions will be almost perfect
result = layer_model.fit(X, y)
assert result.shape == (4,2)
assert np.allclose(result[:,0], y)
assert np.allclose(result[:,1], y)
def test_copy_function_only_model(self):
curLayer = Layer([LinearRegression(), LogisticRegression()])
X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
y = np.dot(X, np.array([1, 2])) + 3
curLayer.fit(X, y)
curLayer2 = curLayer.copy()
gotError = False
try:
curLayer2.predict(np.array([[3, 5]]))
except (NotFittedError):
gotError = True
assert gotError, "Model failed the copy Test: When copying, a deep copy was produced"
def test_stack_copy_function_only_model(self):
first_layer = Layer([LinearRegression(), LogisticRegression()])
second_layer = Layer([LinearRegression()])
model = Stack([first_layer, second_layer])
X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
y = np.dot(X, np.array([1, 2])) + 3
model.fit(X, y)
model2 = model.copy()
gotError = False
try:
model2.predict([1, 2])
except (NotFittedError):
gotError = True
assert gotError, "Model failed the copy Test: When copying, a deep copy was produced"
import pytest
import numpy as np
from sklearn.linear_model import LinearRegression
from sklearn.preprocessing import MinMaxScaler
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import KFold
from picknmix import Layer, Stack
layer_width2_reg = Layer([LinearRegression(), LinearRegression()],
preprocessors = [None, MinMaxScaler()])
layer_width2_clf = Layer([LogisticRegression(solver='liblinear'),
LogisticRegression(solver='liblinear')],
proba=[True,False])
layer_width1_reg = Layer([LinearRegression()])
layer_width1_clf = Layer([LogisticRegression(solver='liblinear')])
class TestStack:
def test_fit_predict_1_layer_reg(self):
model = Stack([layer_width1_reg])
X = np.array([[1, 1], [1, 2], [2, 2], [2, 3]])
y = np.dot(X, np.array([1, 2])) + 3
model.fit(X, y)
result = model.predict(np.array([[3, 5],[3, 5]]))
assert result.shape == (2,)
assert np.allclose(result, np.array([16, 16]))
def test_fit_predict_1_layer_clf(self):
model = Stack([layer_width1_clf])
X = np.array([[1, 1], [1, 1], [0, 0], [0, 0]])
y = np.array([1, 1, 0, 0])
def test_different_numbers_of_preprocessor_and_models(self):
with pytest.raises(Exception):
assert Layer(
[LinearRegression(), LinearRegression()], [MinMaxScaler()])
