def test_formula():
for j in 0, 1:
np.random.seed(34234)
n = 200
y = np.random.randint(0, 2, size=n)
x1 = np.random.normal(size=n)
x2 = np.random.normal(size=n)
g = np.random.randint(0, 25, size=n)
x = np.hstack((x1[:, None], x2[:, None]))
if j == 0:
model1 = ConditionalLogit(y, x, groups=g)
else:
model1 = ConditionalPoisson(y, x, groups=g)
result1 = model1.fit()
df = pd.DataFrame({"y": y, "x1": x1, "x2": x2, "g": g})
if j == 0:
model2 = ConditionalLogit.from_formula(
"y ~ 0 + x1 + x2", groups="g", data=df)
else:
model2 = ConditionalPoisson.from_formula(
"y ~ 0 + x1 + x2", groups="g", data=df)
result2 = model2.fit()
assert_allclose(result1.params, result2.params, rtol=1e-5)
assert_allclose(result1.bse, result2.bse, rtol=1e-5)
assert_allclose(result1.cov_params(), result2.cov_params(), rtol=1e-5)
assert_allclose(result1.tvalues, result2.tvalues, rtol=1e-5)
def test_logit_1d():
y = np.r_[0, 1, 0, 1, 0, 1, 0, 1, 1, 1]
g = np.r_[0, 0, 0, 1, 1, 1, 2, 2, 2, 2]
x = np.r_[0, 1, 0, 0, 1, 1, 0, 0, 1, 0]
x = x[:, None]
model = ConditionalLogit(y, x, groups=g)
# Check the gradient for the denominator of the partial likelihood
for x in -1, 0, 1, 2:
params = np.r_[x, ]
_, grad = model._denom_grad(0, params)
ngrad = approx_fprime(params, lambda x: model._denom(0, x))
assert_allclose(grad, ngrad)
# Check the gradient for the loglikelihood
for x in -1, 0, 1, 2:
grad = approx_fprime(np.r_[x, ], model.loglike)
score = model.score(np.r_[x, ])
assert_allclose(grad, score, rtol=1e-4)
result = model.fit()
# From Stata
assert_allclose(result.params, np.r_[0.9272407], rtol=1e-5)
assert_allclose(result.bse, np.r_[1.295155], rtol=1e-5)
def test_formula():
for j in 0, 1:
np.random.seed(34234)
n = 200
y = np.random.randint(0, 2, size=n)
x1 = np.random.normal(size=n)
x2 = np.random.normal(size=n)
g = np.random.randint(0, 25, size=n)
x = np.hstack((x1[:, None], x2[:, None]))
if j == 0:
model1 = ConditionalLogit(y, x, groups=g)
else:
model1 = ConditionalPoisson(y, x, groups=g)
result1 = model1.fit()
df = pd.DataFrame({"y": y, "x1": x1, "x2": x2, "g": g})
if j == 0:
model2 = ConditionalLogit.from_formula(
"y ~ 0 + x1 + x2", groups="g", data=df)
else:
model2 = ConditionalPoisson.from_formula(
"y ~ 0 + x1 + x2", groups="g", data=df)
result2 = model2.fit()
assert_allclose(result1.params, result2.params, rtol=1e-5)
assert_allclose(result1.bse, result2.bse, rtol=1e-5)
assert_allclose(result1.cov_params(), result2.cov_params(), rtol=1e-5)
assert_allclose(result1.tvalues, result2.tvalues, rtol=1e-5)
def test_logit_2d():
y = np.r_[0, 1, 0, 1, 0, 1, 0, 1, 1, 1]
g = np.r_[0, 0, 0, 1, 1, 1, 2, 2, 2, 2]
x1 = np.r_[0, 1, 0, 0, 1, 1, 0, 0, 1, 0]
x2 = np.r_[0, 0, 1, 0, 0, 1, 0, 1, 1, 1]
x = np.empty((10, 2))
x[:, 0] = x1
x[:, 1] = x2
model = ConditionalLogit(y, x, groups=g)
# Check the gradient for the denominator of the partial likelihood
for x in -1, 0, 1, 2:
params = np.r_[x, -1.5*x]
_, grad = model._denom_grad(0, params)
ngrad = approx_fprime(params, lambda x: model._denom(0, x))
assert_allclose(grad, ngrad, rtol=1e-5)
# Check the gradient for the loglikelihood
for x in -1, 0, 1, 2:
params = np.r_[-0.5*x, 0.5*x]
grad = approx_fprime(params, model.loglike)
score = model.score(params)
assert_allclose(grad, score, rtol=1e-4)
result = model.fit()
# From Stata
assert_allclose(result.params, np.r_[1.011074, 1.236758], rtol=1e-3)
assert_allclose(result.bse, np.r_[1.420784, 1.361738], rtol=1e-5)
result.summary()
def conditional_logit_regression(x, y, groups, method="newton"):
if isinstance(groups, str):
groups = x[groups]
model = ConditionalLogit(y, x, groups=groups)
result = model.fit(method=method)
summary = result.summary()
odds_radio = get_odds_radio(result)
return result, summary, odds_radio
def test_lasso_logistic():
np.random.seed(3423948)
n = 200
groups = np.arange(10)
groups = np.kron(groups, np.ones(n // 10))
group_effects = np.random.normal(size=10)
group_effects = np.kron(group_effects, np.ones(n // 10))
x = np.random.normal(size=(n, 4))
params = np.r_[0, 0, 1, 0]
lin_pred = np.dot(x, params) + group_effects
mean = 1 / (1 + np.exp(-lin_pred))
y = (np.random.uniform(size=n) < mean).astype(np.int)
model0 = ConditionalLogit(y, x, groups=groups)
result0 = model0.fit()
# Should be the same as model0
model1 = ConditionalLogit(y, x, groups=groups)
result1 = model1.fit_regularized(L1_wt=0, alpha=0)
assert_allclose(result0.params, result1.params, rtol=1e-3)
model2 = ConditionalLogit(y, x, groups=groups)
result2 = model2.fit_regularized(L1_wt=1, alpha=0.05)
# Rxegression test
assert_allclose(result2.params, np.r_[0, 0, 0.55235152, 0], rtol=1e-4)
# Test with formula
df = pd.DataFrame({
"y": y,
"x1": x[:, 0],
"x2": x[:, 1],
"x3": x[:, 2],
"x4": x[:, 3],
"groups": groups
})
fml = "y ~ 0 + x1 + x2 + x3 + x4"
model3 = ConditionalLogit.from_formula(fml, groups="groups", data=df)
result3 = model3.fit_regularized(L1_wt=1, alpha=0.05)
assert_allclose(result2.params, result3.params)
def test_logit_1d():
y = np.r_[0, 1, 0, 1, 0, 1, 0, 1, 1, 1]
g = np.r_[0, 0, 0, 1, 1, 1, 2, 2, 2, 2]
x = np.r_[0, 1, 0, 0, 1, 1, 0, 0, 1, 0]
x = x[:, None]
model = ConditionalLogit(y, x, groups=g)
# Check the gradient for the denominator of the partial likelihood
for x in -1, 0, 1, 2:
params = np.r_[x, ]
_, grad = model._denom_grad(0, params)
ngrad = approx_fprime(params, lambda x: model._denom(0, x))
assert_allclose(grad, ngrad)
# Check the gradient for the loglikelihood
for x in -1, 0, 1, 2:
grad = approx_fprime(np.r_[x, ], model.loglike)
score = model.score(np.r_[x, ])
assert_allclose(grad, score, rtol=1e-4)
result = model.fit()
# From Stata
assert_allclose(result.params, np.r_[0.9272407], rtol=1e-5)
assert_allclose(result.bse, np.r_[1.295155], rtol=1e-5)
def test_logit_2d():
y = np.r_[0, 1, 0, 1, 0, 1, 0, 1, 1, 1]
g = np.r_[0, 0, 0, 1, 1, 1, 2, 2, 2, 2]
x1 = np.r_[0, 1, 0, 0, 1, 1, 0, 0, 1, 0]
x2 = np.r_[0, 0, 1, 0, 0, 1, 0, 1, 1, 1]
x = np.empty((10, 2))
x[:, 0] = x1
x[:, 1] = x2
model = ConditionalLogit(y, x, groups=g)
# Check the gradient for the denominator of the partial likelihood
for x in -1, 0, 1, 2:
params = np.r_[x, -1.5*x]
_, grad = model._denom_grad(0, params)
ngrad = approx_fprime(params, lambda x: model._denom(0, x))
assert_allclose(grad, ngrad, rtol=1e-5)
# Check the gradient for the loglikelihood
for x in -1, 0, 1, 2:
params = np.r_[-0.5*x, 0.5*x]
grad = approx_fprime(params, model.loglike)
score = model.score(params)
assert_allclose(grad, score, rtol=1e-4)
result = model.fit()
# From Stata
assert_allclose(result.params, np.r_[1.011074, 1.236758], rtol=1e-3)
assert_allclose(result.bse, np.r_[1.420784, 1.361738], rtol=1e-5)
result.summary()
def test_lasso_logistic():
np.random.seed(3423948)
n = 200
groups = np.arange(10)
groups = np.kron(groups, np.ones(n // 10))
group_effects = np.random.normal(size=10)
group_effects = np.kron(group_effects, np.ones(n // 10))
x = np.random.normal(size=(n, 4))
params = np.r_[0, 0, 1, 0]
lin_pred = np.dot(x, params) + group_effects
mean = 1 / (1 + np.exp(-lin_pred))
y = (np.random.uniform(size=n) < mean).astype(np.int)
model0 = ConditionalLogit(y, x, groups=groups)
result0 = model0.fit()
# Should be the same as model0
model1 = ConditionalLogit(y, x, groups=groups)
result1 = model1.fit_regularized(L1_wt=0, alpha=0)
assert_allclose(result0.params, result1.params, rtol=1e-3)
model2 = ConditionalLogit(y, x, groups=groups)
result2 = model2.fit_regularized(L1_wt=1, alpha=0.05)
# Rxegression test
assert_allclose(result2.params, np.r_[0, 0, 0.55235152, 0], rtol=1e-4)
# Test with formula
df = pd.DataFrame({"y": y, "x1": x[:, 0], "x2": x[:, 1], "x3": x[:, 2],
"x4": x[:, 3], "groups": groups})
fml = "y ~ 0 + x1 + x2 + x3 + x4"
model3 = ConditionalLogit.from_formula(fml, groups="groups", data=df)
result3 = model3.fit_regularized(L1_wt=1, alpha=0.05)
assert_allclose(result2.params, result3.params)
temp = {'是': 1, "否": 0}
for weizhi, f in enumerate(['降低生活水平', '付更高的税']):
temp2 = [0, 0, 0]
temp2[0] = zhi
temp2[1] = temp[data_temp[f]]
temp2[2] = weizhi
tmp = tmp.append(
pd.DataFrame([{
'记录': temp2[0],
'y': temp2[1],
"x": temp2[2]
}]))
tmp = tmp.reset_index()
del tmp['index']
# tmp.to_csv(r"D:/书籍资料整理/属性数据分析/环保_展开.csv")
x = np.array(tmp['x'])
xx = x[:, None]
yy = np.array(tmp['y'])
g = np.array(tmp['记录'])
model = ConditionalLogit(yy, xx, groups=g)
#这里fit的时候会提示删除了905个没有组内方差的.也就是组内数据相同的值.
result = model.fit()
result.summary()
#书中给出的结果为23%,由于书中直接用n求得.
#而用软件为21%。
#R与这个软件一直所以不研究了
# (3)对称logistic 没有找到实现方式并且网上关于这个的文章很少.