def __init__(self):
np.random.seed(54321)
self.endog_n_ = np.random.uniform(0,20,size=30)
self.endog_n_one = self.endog_n_[:,None]
self.exog_n_ = np.random.uniform(0,20,size=30)
self.exog_n_one = self.exog_n_[:,None]
self.degen_exog = self.exog_n_one[:-1]
self.mod1 = OLS(self.endog_n_one, self.exog_n_one)
self.mod1.df_model += 1
#self.mod1.df_resid -= 1
self.res1 = self.mod1.fit()
# Note that these are created for every subclass..
# A little extra overhead probably
self.mod2 = OLS(self.endog_n_one, self.exog_n_one)
self.mod2.df_model += 1
self.res2 = self.mod2.fit()
def __init__(self):
from scikits.statsmodels.datasets.longley import load
data = load()
data.exog = add_constant(data.exog)
self.res1 = OLS(data.endog, data.exog).fit()
R = np.identity(7)[:-1,:]
self.Ftest = self.res1.f_test(R)
class TestDataDimensions(CheckRegressionResults):
def __init__(self):
np.random.seed(54321)
self.endog_n_ = np.random.uniform(0,20,size=30)
self.endog_n_one = self.endog_n_[:,None]
self.exog_n_ = np.random.uniform(0,20,size=30)
self.exog_n_one = self.exog_n_[:,None]
self.degen_exog = self.exog_n_one[:-1]
self.mod1 = OLS(self.endog_n_one, self.exog_n_one)
self.mod1.df_model += 1
#self.mod1.df_resid -= 1
self.res1 = self.mod1.fit()
# Note that these are created for every subclass..
# A little extra overhead probably
self.mod2 = OLS(self.endog_n_one, self.exog_n_one)
self.mod2.df_model += 1
self.res2 = self.mod2.fit()
def check_confidenceintervals(self, conf1, conf2):
assert_almost_equal(conf1, conf2(), DECIMAL_4)
def __init__(self):
from scikits.statsmodels.datasets.longley import load
data = load()
self.data = data
data.exog = add_constant(data.exog)
self.res1 = OLS(data.endog, data.exog).fit()
# def setup(self):
# if skipR:
# raise SkipTest, "Rpy not installed"
# else:
R = np.identity(7)
self.Ttest = self.res1.t_test(R)
class TestTtest(object):
'''
Test individual t-tests. Ie., are the coefficients significantly
different than zero.
'''
def __init__(self):
from scikits.statsmodels.datasets.longley import load
data = load()
self.data = data
data.exog = add_constant(data.exog)
self.res1 = OLS(data.endog, data.exog).fit()
# def setup(self):
# if skipR:
# raise SkipTest, "Rpy not installed"
# else:
R = np.identity(7)
self.Ttest = self.res1.t_test(R)
# self.R_Results = RModel(data.endog, data.exog, r.lm).robj
def test_tvalue(self):
assert_almost_equal(self.Ttest.tvalue, self.res1.t(), DECIMAL_4)
def test_sd(self):
assert_almost_equal(self.Ttest.sd, self.res1.bse, DECIMAL_4)
def test_pvalue(self):
assert_almost_equal(self.Ttest.pvalue,
student_t.sf(np.abs(self.res1.t()),self.res1.model.df_resid),
DECIMAL_4)
def test_df_denom(self):
assert_equal(self.Ttest.df_denom, self.res1.model.df_resid)
def test_effect(self):
assert_almost_equal(self.Ttest.effect, self.res1.params)
class TestFtest(object):
"""
Tests f_test vs. RegressionResults
"""
def __init__(self):
from scikits.statsmodels.datasets.longley import load
data = load()
data.exog = add_constant(data.exog)
self.res1 = OLS(data.endog, data.exog).fit()
R = np.identity(7)[:-1,:]
self.Ftest = self.res1.f_test(R)
def test_F(self):
assert_almost_equal(self.Ftest.fvalue, self.res1.fvalue, DECIMAL_4)
def test_p(self):
assert_almost_equal(self.Ftest.pvalue, self.res1.f_pvalue, DECIMAL_4)
def test_Df_denom(self):
assert_equal(self.Ftest.df_denom, self.res1.model.df_resid)
def test_Df_num(self):
assert_equal(self.Ftest.df_num, 6)
def __init__(self):
super(TestNxNxOne, self).__init__()
self.mod2 = OLS(self.endog_n_, self.exog_n_one)
self.mod2.df_model += 1
self.res2 = self.mod2.fit()
class TestNxNxOne(TestDataDimensions):
def __init__(self):
super(TestNxNxOne, self).__init__()
self.mod2 = OLS(self.endog_n_, self.exog_n_one)
self.mod2.df_model += 1
self.res2 = self.mod2.fit()