def test_crossed_logit_vb_formula():
data = gen_crossed_logit_pandas(10, 10, 1, 2)
fml = "y ~ fe"
fml_vc = {"a": "0 + C(a)", "b": "0 + C(b)"}
glmm1 = BinomialBayesMixedGLM.from_formula(fml, fml_vc, data, vcp_p=0.5)
rslt1 = glmm1.fit_vb()
glmm2 = BinomialBayesMixedGLM(glmm1.endog,
glmm1.exog,
glmm1.exog_vc,
glmm1.ident,
vcp_p=0.5)
rslt2 = glmm2.fit_vb()
assert_allclose(rslt1.params, rslt2.params, atol=1e-4)
rslt1.summary()
rslt2.summary()
for rslt in rslt1, rslt2:
cp = rslt.cov_params()
p = len(rslt.params)
if rslt is rslt1:
assert_equal(cp.shape, np.r_[p, ])
assert_equal(cp > 0, True * np.ones(p))
else:
assert_equal(cp.shape, np.r_[p, ])
assert_equal(cp > 0, True * np.ones(p))
def test_crossed_logit_vb_formula():
data = gen_crossed_logit_pandas(10, 10, 1, 2)
fml = "y ~ fe"
fml_vc = {"a": "0 + C(a)", "b": "0 + C(b)"}
glmm1 = BinomialBayesMixedGLM.from_formula(fml, fml_vc, data, vcp_p=0.5)
rslt1 = glmm1.fit_vb()
glmm2 = BinomialBayesMixedGLM(
glmm1.endog, glmm1.exog, glmm1.exog_vc, glmm1.ident, vcp_p=0.5)
rslt2 = glmm2.fit_vb()
assert_allclose(rslt1.params, rslt2.params, atol=1e-4)
rslt1.summary()
rslt2.summary()
for rslt in rslt1, rslt2:
cp = rslt.cov_params()
p = len(rslt.params)
if rslt is rslt1:
assert_equal(cp.shape, np.r_[p,])
assert_equal(cp > 0, True*np.ones(p))
else:
assert_equal(cp.shape, np.r_[p,])
assert_equal(cp > 0, True*np.ones(p))
def test_crossed_logit_vb():
y, exog_fe, exog_vc, ident = gen_crossed_logit(10, 10, 1, 2)
glmm1 = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident, vcp_p=0.5,
fe_p=0.5)
rslt1 = glmm1.fit_map()
glmm2 = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident, vcp_p=0.5,
fe_p=0.5)
rslt2 = glmm2.fit_vb(mean=rslt1.params)
rslt1.summary()
rslt2.summary()
assert_allclose(rslt1.params[0:5], np.r_[
-5.43073978e-01, -2.46197518e+00, -2.36582801e+00,
-9.64030461e-03, 2.32701078e-03],
rtol=1e-4, atol=1e-4)
assert_allclose(rslt1.cov_params.flat[0:5], np.r_[
4.12927123e-02, -2.04448923e-04, 4.64829219e-05,
1.20377543e-04, -1.45003234e-04],
rtol=1e-4, atol=1e-4)
assert_allclose(rslt2.params[0:5], np.r_[
-0.70834417, -0.3571011, 0.19126823, -0.36074489, 0.058976],
rtol=1e-4, atol=1e-4)
assert_allclose(rslt2.cov_params[0:5], np.r_[
0.05212492, 0.04729656, 0.03916944, 0.25921842, 0.25782576],
rtol=1e-4, atol=1e-4)
def test_elbo_grad():
for f in range(2):
for j in range(2):
if f == 0:
if j == 0:
y, exog_fe, exog_vc, ident = gen_simple_logit(10, 10, 2)
else:
y, exog_fe, exog_vc, ident = gen_crossed_logit(
10, 10, 1, 2)
elif f == 1:
if j == 0:
y, exog_fe, exog_vc, ident = gen_simple_poisson(
10, 10, 0.5)
else:
y, exog_fe, exog_vc, ident = gen_crossed_poisson(
10, 10, 1, 0.5)
exog_vc = sparse.csr_matrix(exog_vc)
if f == 0:
glmm1 = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident,
vcp_p=0.5)
else:
glmm1 = PoissonBayesMixedGLM(y, exog_fe, exog_vc, ident,
vcp_p=0.5)
rslt1 = glmm1.fit_map()
for k in range(3):
if k == 0:
vb_mean = rslt1.params
vb_sd = np.ones_like(vb_mean)
elif k == 1:
vb_mean = np.zeros(len(vb_mean))
vb_sd = np.ones_like(vb_mean)
else:
vb_mean = np.random.normal(size=len(vb_mean))
vb_sd = np.random.uniform(1, 2, size=len(vb_mean))
mean_grad, sd_grad = glmm1.vb_elbo_grad(vb_mean, vb_sd)
def elbo(vec):
n = len(vec) // 2
return glmm1.vb_elbo(vec[:n], vec[n:])
x = np.concatenate((vb_mean, vb_sd))
g1 = approx_fprime(x, elbo, 1e-5)
n = len(x) // 2
mean_grad_n = g1[:n]
sd_grad_n = g1[n:]
assert_allclose(mean_grad, mean_grad_n, atol=1e-2,
rtol=1e-2)
assert_allclose(sd_grad, sd_grad_n, atol=1e-2,
rtol=1e-2)
def test_elbo_grad():
for f in range(2):
for j in range(2):
if f == 0:
if j == 0:
y, exog_fe, exog_vc, ident = gen_simple_logit(10, 10, 2)
else:
y, exog_fe, exog_vc, ident = gen_crossed_logit(
10, 10, 1, 2)
elif f == 1:
if j == 0:
y, exog_fe, exog_vc, ident = gen_simple_poisson(
10, 10, 0.5)
else:
y, exog_fe, exog_vc, ident = gen_crossed_poisson(
10, 10, 1, 0.5)
exog_vc = sparse.csr_matrix(exog_vc)
if f == 0:
glmm1 = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident,
vcp_p=0.5)
else:
glmm1 = PoissonBayesMixedGLM(y, exog_fe, exog_vc, ident,
vcp_p=0.5)
rslt1 = glmm1.fit_map()
for k in range(3):
if k == 0:
vb_mean = rslt1.params
vb_sd = np.ones_like(vb_mean)
elif k == 1:
vb_mean = np.zeros(len(vb_mean))
vb_sd = np.ones_like(vb_mean)
else:
vb_mean = np.random.normal(size=len(vb_mean))
vb_sd = np.random.uniform(1, 2, size=len(vb_mean))
mean_grad, sd_grad = glmm1.vb_elbo_grad(vb_mean, vb_sd)
def elbo(vec):
n = len(vec) // 2
return glmm1.vb_elbo(vec[:n], vec[n:])
x = np.concatenate((vb_mean, vb_sd))
g1 = approx_fprime(x, elbo, 1e-5)
n = len(x) // 2
mean_grad_n = g1[:n]
sd_grad_n = g1[n:]
assert_allclose(mean_grad, mean_grad_n, atol=1e-2,
rtol=1e-2)
assert_allclose(sd_grad, sd_grad_n, atol=1e-2,
rtol=1e-2)
def test_crossed_logit_map():
y, exog_fe, exog_vc, ident = gen_crossed_logit(10, 10, 1, 2)
exog_vc = sparse.csr_matrix(exog_vc)
glmm = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident, vcp_p=0.5)
rslt = glmm.fit_map()
assert_allclose(glmm.logposterior_grad(rslt.params),
np.zeros_like(rslt.params),
atol=1e-4)
def test_crossed_logit_map():
y, exog_fe, exog_vc, ident = gen_crossed_logit(10, 10, 1, 2)
exog_vc = sparse.csr_matrix(exog_vc)
glmm = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident,
vcp_p=0.5)
rslt = glmm.fit_map()
assert_allclose(glmm.logposterior_grad(rslt.params),
np.zeros_like(rslt.params), atol=1e-4)
def test_crossed_logit_vb():
y, exog_fe, exog_vc, ident = gen_crossed_logit(10, 10, 1, 2)
glmm1 = BinomialBayesMixedGLM(
y, exog_fe, exog_vc, ident, vcp_p=0.5, fe_p=0.5)
rslt1 = glmm1.fit_map()
glmm2 = BinomialBayesMixedGLM(
y, exog_fe, exog_vc, ident, vcp_p=0.5, fe_p=0.5)
rslt2 = glmm2.fit_vb(mean=rslt1.params)
rslt1.summary()
rslt2.summary()
assert_allclose(
rslt1.params[0:5],
np.r_[-5.43073978e-01, -2.46197518e+00, -2.36582801e+00,
-9.64030461e-03, 2.32701078e-03],
rtol=1e-4,
atol=1e-4)
assert_allclose(
rslt1.cov_params().flat[0:5],
np.r_[4.12927123e-02, -2.04448923e-04, 4.64829219e-05, 1.20377543e-04,
-1.45003234e-04],
rtol=1e-4,
atol=1e-4)
assert_allclose(
rslt2.params[0:5],
np.r_[-0.70834417, -0.3571011, 0.19126823, -0.36074489, 0.058976],
rtol=1e-4,
atol=1e-4)
assert_allclose(
rslt2.cov_params()[0:5],
np.r_[0.05212492, 0.04729656, 0.03916944, 0.25921842, 0.25782576],
rtol=1e-4,
atol=1e-4)
for rslt in rslt1, rslt2:
cp = rslt.cov_params()
p = len(rslt.params)
if rslt is rslt1:
assert_equal(cp.shape, np.r_[p, p])
np.linalg.cholesky(cp)
else:
assert_equal(cp.shape, np.r_[p,])
assert_equal(cp > 0, True*np.ones(p))
def test_scale_map():
y, exog_fe, exog_vc, ident = gen_simple_logit(10, 10, 0)
exog_fe -= exog_fe.mean(0)
exog_fe /= exog_fe.std(0)
exog_vc = sparse.csr_matrix(exog_vc)
rslts = []
for scale_fe in False, True:
glmm = BinomialBayesMixedGLM(
y, exog_fe, exog_vc, ident, vcp_p=0.5, fe_p=0.5)
rslt = glmm.fit_map(scale_fe=scale_fe)
rslts.append(rslt)
assert_allclose(rslts[0].params, rslts[1].params, rtol=1e-4)
def test_scale_map():
y, exog_fe, exog_vc, ident = gen_simple_logit(10, 10, 0)
exog_fe -= exog_fe.mean(0)
exog_fe /= exog_fe.std(0)
exog_vc = sparse.csr_matrix(exog_vc)
rslts = []
for scale_fe in False, True:
glmm = BinomialBayesMixedGLM(
y, exog_fe, exog_vc, ident, vcp_p=0.5, fe_p=0.5)
rslt = glmm.fit_map(scale_fe=scale_fe)
rslts.append(rslt)
assert_allclose(rslts[0].params, rslts[1].params, rtol=1e-4)
def test_doc_examples():
np.random.seed(8767)
n = 200
m = 20
data = pd.DataFrame({"Year": np.random.uniform(0, 1, n),
"Village": np.random.randint(0, m, n)})
data['year_cen'] = data['Year'] - data.Year.mean()
# Binomial outcome
lpr = np.random.normal(size=m)[data.Village]
lpr += np.random.normal(size=m)[data.Village] * data.year_cen
y = (np.random.uniform(size=n) < 1 / (1 + np.exp(-lpr)))
data["y"] = y.astype(np.int)
# These lines should agree with the example in the class docstring.
random = {"a": '0 + C(Village)', "b": '0 + C(Village)*year_cen'}
model = BinomialBayesMixedGLM.from_formula(
'y ~ year_cen', random, data)
result = model.fit_vb()
_ = result
# Poisson outcome
lpr = np.random.normal(size=m)[data.Village]
lpr += np.random.normal(size=m)[data.Village] * data.year_cen
data["y"] = np.random.poisson(np.exp(lpr))
# These lines should agree with the example in the class docstring.
random = {"a": '0 + C(Village)', "b": '0 + C(Village)*year_cen'}
model = PoissonBayesMixedGLM.from_formula(
'y ~ year_cen', random, data)
result = model.fit_vb()
_ = result
def test_crossed_logit_vb_formula():
data = gen_crossed_logit_pandas(10, 10, 1, 2)
fml = "y ~ fe"
fml_vc = {"a": "0 + C(a)", "b": "0 + C(b)"}
glmm1 = BinomialBayesMixedGLM.from_formula(fml, fml_vc, data, vcp_p=0.5)
rslt1 = glmm1.fit_vb()
glmm2 = BinomialBayesMixedGLM(
glmm1.endog, glmm1.exog, glmm1.exog_vc, glmm1.ident, vcp_p=0.5)
rslt2 = glmm2.fit_vb()
assert_allclose(rslt1.params, rslt2.params, atol=1e-4)
rslt1.summary()
rslt2.summary()
def fit_mixed_lm(subjects_data, formula, random_factors_formulas=None):
model = BinomialBayesMixedGLM.from_formula(formula,
random_factors_formulas,
subjects_data)
result = model.fit_vb()
return model, result
def test_crossed_logit_map():
y, exog_fe, exog_vc, ident = gen_crossed_logit(10, 10, 1, 2)
exog_vc = sparse.csr_matrix(exog_vc)
glmm = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident, vcp_p=0.5)
rslt = glmm.fit_map()
assert_allclose(glmm.logposterior_grad(rslt.params),
np.zeros_like(rslt.params),
atol=1e-4)
# Check dimensions and PSD status of cov_params
cp = rslt.cov_params()
p = len(rslt.params)
assert_equal(cp.shape, np.r_[p, p])
np.linalg.cholesky(cp)
def _train(self, X, y):
# Initialize the output
mapping = {}
# Estimate target type, if necessary
if self.binomial_target is None:
if len(y.unique()) <= 2:
binomial_target = True
else:
binomial_target = False
else:
binomial_target = self.binomial_target
# The estimation does not have to converge -> at least converge to the same value.
np.random.seed(2001)
for switch in self.ordinal_encoder.category_mapping:
col = switch.get('col')
values = switch.get('mapping')
data = self._rename_and_merge(X, y, col)
try:
with warnings.catch_warnings():
warnings.filterwarnings("ignore")
if binomial_target:
# Classification, returns (regularized) log odds per category as stored in vc_mean
# Note: md.predict() returns: output = fe_mean + vcp_mean + vc_mean[category]
md = bgmm.from_formula('target ~ 1', {'a': '0 + C(feature)'}, data).fit_vb()
index_names = [int(float(re.sub(r'C\(feature\)\[(\S+)\]', r'\1', index_name))) for index_name in md.model.vc_names]
estimate = pd.Series(md.vc_mean, index=index_names)
else:
# Regression, returns (regularized) mean deviation of the observation's category from the global mean
md = smf.mixedlm('target ~ 1', data, groups=data['feature']).fit()
tmp = dict()
for key, value in md.random_effects.items():
tmp[key] = value[0]
estimate = pd.Series(tmp)
except np.linalg.LinAlgError:
# Singular matrix -> just return all zeros
estimate = pd.Series(np.zeros(len(values)), index=values)
# Ignore unique columns. This helps to prevent overfitting on id-like columns
if len(X[col].unique()) == len(y):
estimate[:] = 0
if self.handle_unknown == 'return_nan':
estimate.loc[-1] = np.nan
elif self.handle_unknown == 'value':
estimate.loc[-1] = 0
if self.handle_missing == 'return_nan':
estimate.loc[values.loc[np.nan]] = np.nan
elif self.handle_missing == 'value':
estimate.loc[-2] = 0
mapping[col] = estimate
return mapping
def test_crossed_logit_map():
y, exog_fe, exog_vc, ident = gen_crossed_logit(10, 10, 1, 2)
exog_vc = sparse.csr_matrix(exog_vc)
glmm = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident, vcp_p=0.5)
rslt = glmm.fit_map()
assert_allclose(
glmm.logposterior_grad(rslt.params),
np.zeros_like(rslt.params),
atol=1e-4)
# Check dimensions and PSD status of cov_params
cp = rslt.cov_params()
p = len(rslt.params)
assert_equal(cp.shape, np.r_[p, p])
np.linalg.cholesky(cp)
def test_crossed_logit_vb_formula():
data = gen_crossed_logit_pandas(10, 10, 1, 2)
fml = "y ~ fe"
fml_vc = {"a": "0 + C(a)", "b": "0 + C(b)"}
glmm1 = BinomialBayesMixedGLM.from_formula(
fml, fml_vc, data, vcp_p=0.5)
rslt1 = glmm1.fit_vb()
glmm2 = BinomialBayesMixedGLM(glmm1.endog, glmm1.exog_fe, glmm1.exog_vc,
glmm1.ident, vcp_p=0.5)
rslt2 = glmm2.fit_vb()
assert_allclose(rslt1.params, rslt2.params, atol=1e-4)
rslt1.summary()
rslt2.summary()
def test_simple_logit_map():
y, exog_fe, exog_vc, ident = gen_simple_logit(10, 10, 2)
exog_vc = sparse.csr_matrix(exog_vc)
glmm = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident, vcp_p=0.5)
rslt = glmm.fit_map()
assert_allclose(glmm.logposterior_grad(rslt.params),
np.zeros_like(rslt.params),
atol=1e-3)
# Test the predict method
for linear in False, True:
for exog in None, exog_fe:
pr1 = rslt.predict(linear=linear, exog=exog)
pr2 = glmm.predict(rslt.params, linear=linear, exog=exog)
assert_allclose(pr1, pr2)
if not linear:
assert_equal(pr1.min() >= 0, True)
assert_equal(pr1.max() <= 1, True)
def test_crossed_logit_vb():
y, exog_fe, exog_vc, ident = gen_crossed_logit(10, 10, 1, 2)
glmm1 = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident, vcp_p=0.5,
fe_p=0.5)
rslt1 = glmm1.fit_map()
glmm2 = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident, vcp_p=0.5,
fe_p=0.5)
rslt2 = glmm2.fit_vb(mean=rslt1.params)
rslt1.summary()
rslt2.summary()
assert_allclose(rslt1.params[0:5], np.r_[
-5.43073978e-01, -2.46197518e+00, -2.36582801e+00,
-9.64030461e-03, 2.32701078e-03],
rtol=1e-4, atol=1e-4)
assert_allclose(rslt1.cov_params.flat[0:5], np.r_[
4.12927123e-02, -2.04448923e-04, 4.64829219e-05,
1.20377543e-04, -1.45003234e-04],
rtol=1e-4, atol=1e-4)
assert_allclose(rslt2.params[0:5], np.r_[
-0.70834417, -0.3571011, 0.19126823, -0.36074489, 0.058976],
rtol=1e-4, atol=1e-4)
assert_allclose(rslt2.cov_params[0:5], np.r_[
0.05212492, 0.04729656, 0.03916944, 0.25921842, 0.25782576],
rtol=1e-4, atol=1e-4)
def test_simple_logit_vb():
y, exog_fe, exog_vc, ident = gen_simple_logit(10, 10, 0)
exog_vc = sparse.csr_matrix(exog_vc)
glmm1 = BinomialBayesMixedGLM(y,
exog_fe,
exog_vc,
ident,
vcp_p=0.5,
fe_p=0.5)
rslt1 = glmm1.fit_map()
glmm2 = BinomialBayesMixedGLM(y,
exog_fe,
exog_vc,
ident,
vcp_p=0.5,
fe_p=0.5)
rslt2 = glmm2.fit_vb(rslt1.params)
rslt1.summary()
rslt2.summary()
assert_allclose(rslt1.params[0:5],
np.r_[0.75330405, -0.71643228, -2.49091288, -0.00959806,
0.00450254],
rtol=1e-4,
atol=1e-4)
assert_allclose(rslt2.params[0:5],
np.r_[0.79338836, -0.7599833, -0.64149356, -0.24772884,
0.10775366],
rtol=1e-4,
atol=1e-4)
def test_simple_logit_map():
y, exog_fe, exog_vc, ident = gen_simple_logit(10, 10, 2)
exog_vc = sparse.csr_matrix(exog_vc)
glmm = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident, vcp_p=0.5)
rslt = glmm.fit_map()
assert_allclose(
glmm.logposterior_grad(rslt.params),
np.zeros_like(rslt.params),
atol=1e-3)
# Test the predict method
for linear in False, True:
for exog in None, exog_fe:
pr1 = rslt.predict(linear=linear, exog=exog)
pr2 = glmm.predict(rslt.params, linear=linear, exog=exog)
assert_allclose(pr1, pr2)
if not linear:
assert_equal(pr1.min() >= 0, True)
assert_equal(pr1.max() <= 1, True)
def test_simple_logit_vb():
y, exog_fe, exog_vc, ident = gen_simple_logit(10, 10, 0)
exog_vc = sparse.csr_matrix(exog_vc)
glmm1 = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident, vcp_p=0.5,
fe_p=0.5)
rslt1 = glmm1.fit_map()
glmm2 = BinomialBayesMixedGLM(y, exog_fe, exog_vc, ident, vcp_p=0.5,
fe_p=0.5)
rslt2 = glmm2.fit_vb(rslt1.params)
rslt1.summary()
rslt2.summary()
assert_allclose(rslt1.params[0:5], np.r_[
0.75330405, -0.71643228, -2.49091288, -0.00959806, 0.00450254],
rtol=1e-4, atol=1e-4)
assert_allclose(rslt2.params[0:5], np.r_[
0.79338836, -0.7599833, -0.64149356, -0.24772884, 0.10775366],
rtol=1e-4, atol=1e-4)
def test_simple_logit_vb():
y, exog_fe, exog_vc, ident = gen_simple_logit(10, 10, 0)
exog_vc = sparse.csr_matrix(exog_vc)
glmm1 = BinomialBayesMixedGLM(
y, exog_fe, exog_vc, ident, vcp_p=0.5, fe_p=0.5)
rslt1 = glmm1.fit_map()
glmm2 = BinomialBayesMixedGLM(
y, exog_fe, exog_vc, ident, vcp_p=0.5, fe_p=0.5)
rslt2 = glmm2.fit_vb(rslt1.params)
rslt1.summary()
rslt2.summary()
assert_allclose(
rslt1.params[0:5],
np.r_[0.75330405, -0.71643228, -2.49091288, -0.00959806, 0.00450254],
rtol=1e-4,
atol=1e-4)
assert_allclose(
rslt2.params[0:5],
np.r_[0.79338836, -0.7599833, -0.64149356, -0.24772884, 0.10775366],
rtol=1e-4,
atol=1e-4)
for rslt in rslt1, rslt2:
cp = rslt.cov_params()
p = len(rslt.params)
if rslt is rslt1:
assert_equal(cp.shape, np.r_[p, p])
np.linalg.cholesky(cp)
else:
assert_equal(cp.shape, np.r_[p,])
assert_equal(cp > 0, True*np.ones(p))
def test_crossed_logit_vb():
y, exog_fe, exog_vc, ident = gen_crossed_logit(10, 10, 1, 2)
glmm1 = BinomialBayesMixedGLM(y,
exog_fe,
exog_vc,
ident,
vcp_p=0.5,
fe_p=0.5)
rslt1 = glmm1.fit_map()
glmm2 = BinomialBayesMixedGLM(y,
exog_fe,
exog_vc,
ident,
vcp_p=0.5,
fe_p=0.5)
rslt2 = glmm2.fit_vb(mean=rslt1.params)
rslt1.summary()
rslt2.summary()
assert_allclose(rslt1.params[0:5],
np.r_[-5.43073978e-01, -2.46197518e+00, -2.36582801e+00,
-9.64030461e-03, 2.32701078e-03],
rtol=1e-4,
atol=1e-4)
assert_allclose(rslt1.cov_params().flat[0:5],
np.r_[4.12927123e-02, -2.04448923e-04, 4.64829219e-05,
1.20377543e-04, -1.45003234e-04],
rtol=1e-4,
atol=1e-4)
assert_allclose(rslt2.params[0:5],
np.r_[-0.70834417, -0.3571011, 0.19126823, -0.36074489,
0.058976],
rtol=1e-4,
atol=1e-4)
assert_allclose(rslt2.cov_params()[0:5],
np.r_[0.05212492, 0.04729656, 0.03916944, 0.25921842,
0.25782576],
rtol=1e-4,
atol=1e-4)
for rslt in rslt1, rslt2:
cp = rslt.cov_params()
p = len(rslt.params)
if rslt is rslt1:
assert_equal(cp.shape, np.r_[p, p])
np.linalg.cholesky(cp)
else:
assert_equal(cp.shape, np.r_[p, ])
assert_equal(cp > 0, True * np.ones(p))
def test_logit_map_crossed_formula():
data = gen_crossed_logit_pandas(10, 10, 1, 0.5)
fml = "y ~ fe"
fml_vc = {"a": "0 + C(a)", "b": "0 + C(b)"}
glmm = BinomialBayesMixedGLM.from_formula(
fml, fml_vc, data, vcp_p=0.5)
rslt = glmm.fit_map()
assert_allclose(glmm.logposterior_grad(rslt.params),
np.zeros_like(rslt.params), atol=1e-4)
rslt.summary()
r = rslt.random_effects("a")
assert_allclose(r.iloc[0, :].values,
np.r_[-0.02004904, 0.094014], atol=1e-4)
def test_logit_map_crossed_formula():
data = gen_crossed_logit_pandas(10, 10, 1, 0.5)
fml = "y ~ fe"
fml_vc = {"a": "0 + C(a)", "b": "0 + C(b)"}
glmm = BinomialBayesMixedGLM.from_formula(
fml, fml_vc, data, vcp_p=0.5)
rslt = glmm.fit_map()
assert_allclose(glmm.logposterior_grad(rslt.params),
np.zeros_like(rslt.params), atol=1e-4)
rslt.summary()
r = rslt.random_effects("a")
assert_allclose(r.iloc[0, :].values,
np.r_[-0.02004904, 0.094014], atol=1e-4)
def test_simple_logit_vb():
y, exog_fe, exog_vc, ident = gen_simple_logit(10, 10, 0)
exog_vc = sparse.csr_matrix(exog_vc)
glmm1 = BinomialBayesMixedGLM(y,
exog_fe,
exog_vc,
ident,
vcp_p=0.5,
fe_p=0.5)
rslt1 = glmm1.fit_map()
glmm2 = BinomialBayesMixedGLM(y,
exog_fe,
exog_vc,
ident,
vcp_p=0.5,
fe_p=0.5)
rslt2 = glmm2.fit_vb(rslt1.params)
rslt1.summary()
rslt2.summary()
assert_allclose(rslt1.params[0:5],
np.r_[0.75330405, -0.71643228, -2.49091288, -0.00959806,
0.00450254],
rtol=1e-4,
atol=1e-4)
assert_allclose(rslt2.params[0:5],
np.r_[0.79338836, -0.7599833, -0.64149356, -0.24772884,
0.10775366],
rtol=1e-4,
atol=1e-4)
for rslt in rslt1, rslt2:
cp = rslt.cov_params()
p = len(rslt.params)
if rslt is rslt1:
assert_equal(cp.shape, np.r_[p, p])
np.linalg.cholesky(cp)
else:
assert_equal(cp.shape, np.r_[p, ])
assert_equal(cp > 0, True * np.ones(p))
def test_logit_map_crossed_formula():
data = gen_crossed_logit_pandas(10, 10, 1, 0.5)
fml = "y ~ fe"
fml_vc = {"a": "0 + C(a)", "b": "0 + C(b)"}
glmm = BinomialBayesMixedGLM.from_formula(fml, fml_vc, data, vcp_p=0.5)
rslt = glmm.fit_map()
assert_allclose(glmm.logposterior_grad(rslt.params),
np.zeros_like(rslt.params),
atol=1e-4)
rslt.summary()
r = rslt.random_effects("a")
assert_allclose(r.iloc[0, :].values,
np.r_[-0.02004904, 0.094014],
atol=1e-4)
# Check dimensions and PSD status of cov_params
cm = rslt.cov_params()
p = rslt.params.shape[0]
assert_equal(list(cm.shape), [p, p])
np.linalg.cholesky(cm)
def test_logit_map_crossed_formula():
data = gen_crossed_logit_pandas(10, 10, 1, 0.5)
fml = "y ~ fe"
fml_vc = {"a": "0 + C(a)", "b": "0 + C(b)"}
glmm = BinomialBayesMixedGLM.from_formula(fml, fml_vc, data, vcp_p=0.5)
rslt = glmm.fit_map()
assert_allclose(
glmm.logposterior_grad(rslt.params),
np.zeros_like(rslt.params),
atol=1e-4)
rslt.summary()
r = rslt.random_effects("a")
assert_allclose(
r.iloc[0, :].values, np.r_[-0.02004904, 0.094014], atol=1e-4)
# Check dimensions and PSD status of cov_params
cm = rslt.cov_params()
p = rslt.params.shape[0]
assert_equal(list(cm.shape), [p, p])
np.linalg.cholesky(cm)
df = get_data(group)
for outcome in "bucketacc", "bucketcomp":
fml = get_formula(adj_time=adj_time)
fmx = outcome + " ~ " + fml
yl = {
"bucketacc": "target accuracy",
"bucketcomp": "competitor accuracy"
}[outcome]
vcx = get_vcf(vcs, adj_time)
fmx = outcome + " ~ " + fml
model = BinomialBayesMixedGLM.from_formula(
fmx, vcx, df, vcp_p=3, fe_p=3)
fid = open(group + ".pkl", "rb")
pars = pickle.load(fid)
fid.close()
if adj_time:
tm = pars["tm_adj"]
ts = pars["ts_adj"]
else:
tm = pars["tm"]
ts = pars["ts"]
if use_vb:
params = pd.read_csv(
"%s_params_%d_%s_%s_vb.csv" % (group, vcs, outcome, adjs))
else:
select_cols = [
"play_id", "game_id", "touchdown", "yards_gained", "turnover", "posteam",
"defteam", "yardline_100", "half_seconds_remaining", "play_type",
"shotgun", "no_huddle", "qb_dropback", "pass_length", "pass_location",
"run_location", "run_gap", "field_goal_result", "opp_fg_prob",
"opp_td_prob", "fumble_forced", "fumble_not_forced", "fumble_lost",
"penalty"
]
nfl_rush_2019 = nfl_2019[nfl_2019["play_type"] == "rush"]
# first, fit rush outcome models
# 1 - touchdown
rush_penalty_mod = BinomialBayesMixedGLM.from_formula(
'penalty ~ shotgun + no_huddle + qb_dropback + run_location + run_gap',
['0 + rusher_id', '0 + def_id'],
data=nfl_rush_2019)
rush_penalty_result = rush_penalty_mod.fit_vb
# 2 - rushing yards
rush_yard_mod = PoissonBayesMixedGLM.from_formula(
'yards_gained ~ shotgun + no_huddle + qb_dropback + run_location + run_gap',
['0 + rusher_id', '0 + def_id'],
data=nfl_rush_2019)
rush_yard_result = rush_yard_mod.fit_vb()
# 3 - rushing turnovers (fumbles)
rush_turnover_mod = BinomialBayesMixedGLM.from_formula(
'turnover ~ shotgun + no_huddle + qb_dropback + run_location + run_gap',
vcn = ["Sample", "Exon", "Gene", "Person"]
fml = "Imprinted ~ KidRank + C(Lib) + Boy"
vc_fml = {"Sample": "0 + C(Sample)", "Exon": "0 + C(Exon)",
"Gene": "0 + C(Gene)", "Person": "0 + C(Person)"}
if kc == 3:
fml += " + Pat"
fml = fml.replace("Pat", "Pat01")
fml = fml.replace("C(Lib)", "C(Lib)*Pat01")
fml = fml.replace("KidRank", "KidRank*Pat01")
dy = dx.drop("PlacentaWeight", axis=1)
if kc != 3:
model = BinomialBayesMixedGLM.from_formula(fml, vc_fml, dy, vcp_p=3, fe_p=3)
else:
ident = []
exog_vc = []
for g in dy.Gene.unique():
ident.append(genecode[g])
exog_vc.append((dy.Gene == g).astype(np.int))
for e in dy.Exon.unique():
ident.append(exoncode[e])
exog_vc.append((dy.Exon == e).astype(np.int))
for p in dy.Person.unique():
ident.append(4)
exog_vc.append((dy.Person == p).astype(np.int))
def glmm_model(data, features, y, random_effects):
model = BinomialBayesMixedGLM.from_formula(f'{y} ~ {features}',
random_effects, data)
result = model.fit_vb()
return result
tmp['诊断严重程度'] = tmp['诊断严重程度'].replace({'轻微': 0, '严重': 1})
tmp['治疗'] = tmp['治疗'].replace({'标准': 0, '新药': 1})
tmp = tmp.reset_index()
del tmp['index']
tmp = tmp.rename(columns={
'周数': 'zhous',
'值': 'zhi',
'组': 'zu',
'诊断严重程度': 'severity',
'治疗': 'drug'
})
tmp.to_csv(r"D:/书籍资料整理/属性数据分析/抑郁症治疗_展开.csv")
random = {"a": '0 + C(zu)'}
model = BinomialBayesMixedGLM.from_formula(
'zhi ~ severity + drug + zhous+drug:zhous', random, tmp)
result = model.fit_vb()
#给出的结果大致上与书中的结果差不多,估计差异在
#书中给出的结果为使用高斯-埃尔米特求积
#而statsmodels使用的是贝叶斯方法.
#结果给出的是方差可能要开根号才能求标准差
#另外
print(result.summary())
data = pd.read_csv(r"D:/书籍资料整理/属性数据分析/老鼠.csv")
random = {"a": '0 + C(簇)'}
model = BinomialBayesMixedGLM.from_formula('死亡 ~ C(组) ', random, data)
result = model.fit_vb()
result.summary()
import numpy as np
from statsmodels.genmod.bayes_mixed_glm import (BinomialBayesMixedGLM,
PoissonBayesMixedGLM)
import pandas as pd
from scipy import sparse
from numpy.testing import assert_allclose, assert_equal
from scipy.optimize import approx_fprime
np.random.seed(8767)
n = 200
m = 20
data = pd.DataFrame({
"Year": np.random.uniform(0, 1, n),
"Village": np.random.randint(0, m, n)
})
data['year_cen'] = data['Year'] - data.Year.mean()
# Binomial outcome
lpr = np.random.normal(size=m)[data.Village]
lpr += np.random.normal(size=m)[data.Village] * data.year_cen
y = (np.random.uniform(size=n) < 1 / (1 + np.exp(-lpr)))
data["y"] = y.astype(int)
# These lines should agree with the example in the class docstring.
random = {"a": '0 + C(Village)'}
print(data)
model = BinomialBayesMixedGLM.from_formula('y ~ year_cen', random, data)
result = model.fit_vb()