def test_OLS_regi(self):
#Artficial:
n = 256
x1 = np.random.uniform(-10, 10, (n, 1))
y = np.dot(np.hstack((np.ones(
(n, 1)), x1)), np.array([[1], [0.5]])) + np.random.normal(
0, 1, (n, 1))
latt = int(np.sqrt(n))
regi = [0] * (n / 2) + [1] * (n / 2)
model = OLS_Regimes(y,
x1,
regimes=regi,
regime_err_sep=True,
sig2n_k=False)
model1 = OLS(y[0:(n / 2)].reshape((n / 2), 1),
x1[0:(n / 2)],
sig2n_k=False)
model2 = OLS(y[(n / 2):n].reshape((n / 2), 1),
x1[(n / 2):n],
sig2n_k=False)
tbetas = np.vstack((model1.betas, model2.betas))
np.testing.assert_array_almost_equal(model.betas, tbetas)
vm = np.hstack((model1.vm.diagonal(), model2.vm.diagonal()))
np.testing.assert_array_almost_equal(model.vm.diagonal(), vm, 6)
#Columbus:
reg = OLS_Regimes(self.y,
self.x,
self.regimes,
w=self.w,
constant_regi='many',
nonspat_diag=True,
spat_diag=True,
name_y=self.y_var,
name_x=self.x_var,
name_ds='columbus',
name_regimes=self.r_var,
name_w='columbus.gal',
regime_err_sep=True)
np.testing.assert_array_almost_equal(reg.multi[0].aic,
192.96044303402897, 7)
tbetas = np.array([[68.78670869], [-1.9864167], [-0.10887962],
[67.73579559], [-1.36937552], [-0.31792362]])
np.testing.assert_array_almost_equal(tbetas, reg.betas)
vm = np.array([41.68828023, -1.83582717, -0.17053478, 0., 0., 0.])
np.testing.assert_array_almost_equal(reg.vm[0], vm, 6)
u_3 = np.array([[0.31781838], [-5.6905584], [-6.8819715]])
np.testing.assert_array_almost_equal(reg.u[0:3], u_3, 7)
predy_3 = np.array([[15.40816162], [24.4923124], [37.5087525]])
np.testing.assert_array_almost_equal(reg.predy[0:3], predy_3, 7)
chow_regi = np.array([[0.01002733,
0.92023592], [0.46017009, 0.49754449],
[0.60732697, 0.43579603]])
np.testing.assert_array_almost_equal(reg.chow.regi, chow_regi, 7)
self.assertAlmostEqual(reg.chow.joint[0], 0.67787986791767096, 7)
def test_OLS(self):
start_suppress = np.get_printoptions()['suppress']
np.set_printoptions(suppress=True)
ols = OLS_Regimes(self.y,
self.x,
self.regimes,
w=self.w,
regime_err_sep=False,
constant_regi='many',
nonspat_diag=False,
spat_diag=True,
name_y=self.y_var,
name_x=self.x_var,
name_ds='columbus',
name_regimes=self.r_var,
name_w='columbus.gal')
#np.testing.assert_array_almost_equal(ols.aic, 408.73548964604873 ,7)
np.testing.assert_array_almost_equal(ols.ar2, 0.50761700679873101, 7)
np.testing.assert_array_almost_equal(ols.betas,np.array([[ 68.78670869],\
[ -1.9864167 ],[ -0.10887962],[ 67.73579559],[ -1.36937552],[ -0.31792362]]))
vm = np.array([48.81339213, -2.14959579, -0.19968157, 0., 0., 0.])
np.testing.assert_array_almost_equal(ols.vm[0], vm, 6)
np.testing.assert_array_almost_equal(ols.lm_error, \
(5.92970357, 0.01488775),7)
np.testing.assert_array_almost_equal(ols.lm_lag, \
(8.78315751, 0.00304024), 7)
np.testing.assert_array_almost_equal(ols.lm_sarma, \
(8.89955982, 0.01168114), 7)
np.testing.assert_array_almost_equal(ols.mean_y, \
35.1288238979591,7)
np.testing.assert_equal(ols.k, 6)
np.testing.assert_equal(ols.kf, 0)
np.testing.assert_equal(ols.kr, 3)
np.testing.assert_equal(ols.n, 49)
np.testing.assert_equal(ols.nr, 2)
np.testing.assert_equal(ols.name_ds, 'columbus')
np.testing.assert_equal(ols.name_gwk, None)
np.testing.assert_equal(ols.name_w, 'columbus.gal')
np.testing.assert_equal(ols.name_x, [
'0_CONSTANT', '0_INC', '0_HOVAL', '1_CONSTANT', '1_INC', '1_HOVAL'
])
np.testing.assert_equal(ols.name_y, 'CRIME')
np.testing.assert_array_almost_equal(ols.predy[3],
np.array([51.05003696]), 7)
np.testing.assert_array_almost_equal(ols.r2, \
0.55890690192386316 ,7)
np.testing.assert_array_almost_equal(ols.rlm_error, \
(0.11640231, 0.73296972),7)
np.testing.assert_array_almost_equal(ols.rlm_lag, \
(2.96985625, 0.08482939), 7)
np.testing.assert_equal(ols.robust, 'unadjusted')
np.testing.assert_array_almost_equal(ols.sig2, \
137.84897351821013,7 )
np.testing.assert_array_almost_equal(ols.sig2n, \
120.96950737312316, 7)
np.testing.assert_array_almost_equal(ols.t_stat[2][0], \
-0.43342216706091791,7)
np.testing.assert_array_almost_equal(ols.t_stat[2][1], \
0.66687472578594531,7)
np.set_printoptions(suppress=start_suppress)
def test_OLS_regime_err_sep(self):
start_suppress = np.get_printoptions()['suppress']
np.set_printoptions(suppress=True)
ols = OLS_Regimes(self.y,
self.x,
self.regimes,
w=self.w,
regime_err_sep=True,
constant_regi='many',
nonspat_diag=False,
spat_diag=True,
name_y=self.y_var,
name_x=self.x_var,
name_ds='columbus',
name_regimes=self.r_var,
name_w='columbus.gal')
np.testing.assert_array_almost_equal(ols.betas,np.array([[ 68.78670869],\
[ -1.9864167 ],[ -0.10887962],[ 67.73579559],[ -1.36937552],[ -0.31792362]]))
vm = np.array([41.68828023, -1.83582717, -0.17053478, 0., 0., 0.])
np.testing.assert_array_almost_equal(ols.vm[0], vm, 6)
np.testing.assert_array_almost_equal(ols.multi[0].lm_error, \
(5.45795541, 0.01947943),7)
np.testing.assert_array_almost_equal(ols.multi[0].lm_lag, \
(8.20573925, 0.00417581), 7)
np.testing.assert_array_almost_equal(ols.multi[0].lm_sarma, \
(8.39852379, 0.01500665), 7)
np.testing.assert_array_almost_equal(ols.multi[0].mean_y, \
31.614447920000003,7)
np.testing.assert_equal(ols.kf, 0)
np.testing.assert_equal(ols.kr, 3)
np.testing.assert_equal(ols.n, 49)
np.testing.assert_equal(ols.nr, 2)
np.testing.assert_equal(ols.name_ds, 'columbus')
np.testing.assert_equal(ols.name_gwk, None)
np.testing.assert_equal(ols.name_w, 'columbus.gal')
np.testing.assert_equal(ols.name_x, [
'0_CONSTANT', '0_INC', '0_HOVAL', '1_CONSTANT', '1_INC', '1_HOVAL'
])
np.testing.assert_equal(ols.multi[0].name_y, '0_CRIME')
np.testing.assert_array_almost_equal(ols.predy[3],
np.array([51.05003696]), 7)
np.testing.assert_array_almost_equal(ols.multi[0].r2, \
0.6298789694928845 ,7)
np.testing.assert_equal(ols.multi[0].robust, 'unadjusted')
np.testing.assert_array_almost_equal(ols.multi[0].sig2, \
117.72766418598968,7 )
np.testing.assert_array_almost_equal(ols.multi[0].sig2n, \
103.60034448367092, 7)
np.testing.assert_array_almost_equal(ols.multi[0].t_stat[2][0], \
-0.46900084351673077,7)
np.testing.assert_array_almost_equal(ols.multi[0].t_stat[2][1], \
0.6436834438232635,7)
np.set_printoptions(suppress=start_suppress)
def test_OLS_fixed(self):
start_suppress = np.get_printoptions()['suppress']
np.set_printoptions(suppress=True)
ols = OLS_Regimes(self.y,
self.x,
self.regimes,
w=self.w,
cols2regi=[False, True],
regime_err_sep=True,
constant_regi='one',
nonspat_diag=False,
spat_diag=True,
name_y=self.y_var,
name_x=self.x_var,
name_ds='columbus',
name_regimes=self.r_var,
name_w='columbus.gal')
np.testing.assert_array_almost_equal(
ols.betas,
np.array([[-0.24385565], [-0.26335026], [68.89701137],
[-1.67389685]]))
vm = np.array([0.02354271, 0.01246677, 0.00424658, -0.04921356])
np.testing.assert_array_almost_equal(ols.vm[0], vm, 6)
np.testing.assert_array_almost_equal(ols.lm_error, \
(5.62668744, 0.01768903),7)
np.testing.assert_array_almost_equal(ols.lm_lag, \
(9.43264957, 0.00213156), 7)
np.testing.assert_array_almost_equal(ols.mean_y, \
35.12882389795919,7)
np.testing.assert_equal(ols.kf, 2)
np.testing.assert_equal(ols.kr, 1)
np.testing.assert_equal(ols.n, 49)
np.testing.assert_equal(ols.nr, 2)
np.testing.assert_equal(ols.name_ds, 'columbus')
np.testing.assert_equal(ols.name_gwk, None)
np.testing.assert_equal(ols.name_w, 'columbus.gal')
np.testing.assert_equal(
ols.name_x,
['0_HOVAL', '1_HOVAL', '_Global_CONSTANT', '_Global_INC'])
np.testing.assert_equal(ols.name_y, 'CRIME')
np.testing.assert_array_almost_equal(ols.predy[3],
np.array([52.65974636]), 7)
np.testing.assert_array_almost_equal(ols.r2, \
0.5525561183786056 ,7)
np.testing.assert_equal(ols.robust, 'unadjusted')
np.testing.assert_array_almost_equal(ols.t_stat[2][0], \
13.848705206463748,7)
np.testing.assert_array_almost_equal(ols.t_stat[2][1], \
7.776650625274256e-18,7)
np.set_printoptions(suppress=start_suppress)