52 lines
1.3 KiB
Python
52 lines
1.3 KiB
Python
# Author: Alexandre Gramfort <alexandre.gramfort@inria.fr>
|
|
# Fabian Pedregosa <fabian.pedregosa@inria.fr>
|
|
#
|
|
# License: BSD Style.
|
|
|
|
from numpy.testing import assert_array_almost_equal
|
|
import numpy as np
|
|
from scipy import sparse
|
|
|
|
from ..base import LinearRegression
|
|
from ...utils import check_random_state
|
|
|
|
|
|
def test_linear_regression():
|
|
"""
|
|
Test LinearRegression on a simple dataset.
|
|
"""
|
|
# a simple dataset
|
|
X = [[1], [2]]
|
|
Y = [1, 2]
|
|
|
|
clf = LinearRegression()
|
|
clf.fit(X, Y)
|
|
|
|
assert_array_almost_equal(clf.coef_, [1])
|
|
assert_array_almost_equal(clf.intercept_, [0])
|
|
assert_array_almost_equal(clf.predict(X), [1, 2])
|
|
|
|
# test it also for degenerate input
|
|
X = [[1]]
|
|
Y = [0]
|
|
|
|
clf = LinearRegression()
|
|
clf.fit(X, Y)
|
|
assert_array_almost_equal(clf.coef_, [0])
|
|
assert_array_almost_equal(clf.intercept_, [0])
|
|
assert_array_almost_equal(clf.predict(X), [0])
|
|
|
|
|
|
def test_linear_regression_sparse(random_state=0):
|
|
"Test that linear regression also works with sparse data"
|
|
random_state = check_random_state(random_state)
|
|
n = 100
|
|
X = sparse.eye(n, n)
|
|
beta = random_state.rand(n)
|
|
y = X * beta[:, np.newaxis]
|
|
|
|
ols = LinearRegression()
|
|
ols.fit(X, y.ravel())
|
|
assert_array_almost_equal(beta, ols.coef_ + ols.intercept_)
|
|
assert_array_almost_equal(ols.residues_, 0)
|