def suggest_burn(self):
"""
Suggest a number of burn-in iterations. For Gaussian models this will
be based on the simulated values of log-likelihood. For other models
it will be a fixed percentage of the draws.
"""
loglike = self.log_likelihood
if loglike is None:
burn = self.niter / 10
else:
burn = R.suggest_burn(loglike)
return burn
def coefficient_positive_probability(self, burn=None):
"""
Args:
burn: The number of MCMC iterations to discard as burn-in.
Returns:
A pd.Series containing the marginal probability that each coefficient
is positive.
"""
if burn is None:
burn = R.suggest_burn(self.log_likelihood)
probs = coefficient_positive_probability(
self._coefficient_draws[burn:, :])
return pd.Series(probs, index=self.xnames)
def inclusion_probs(self, burn=None):
"""
Args:
burn: The number of MCMC iterations to discard as burn-in.
Returns:
A pd.Series containing the marginal probability that each coefficient
is nonzero. The series is indexed by the set of variable names.
"""
if burn is None:
burn = R.suggest_burn(self.log_likelihood)
probs = compute_inclusion_probabilities(
self._coefficient_draws[burn:, ])
return pd.Series(probs, index=self.xnames)
def predict(self, newdata, burn=None, seed=None):
"""
Return an LmSpikePrediciton object.
"""
if burn is None:
burn = R.suggest_burn(self.log_likelihood)
if seed is not None:
boom.GlobalRng.rng.seed(int(seed))
if isinstance(newdata, np.ndarray) and len(newdata.shape) == 1:
newdata = newdata.reshape(1, -1)
if isinstance(newdata, np.ndarray) and newdata.shape[1] == self.xdim:
predictors = newdata
else:
predictors = patsy.build_design_matrices([self._x_design_info],
data=newdata)[0]
return self._coefficient_draws[burn:, :] @ predictors.T
def coefficient_positive_probability(self, burn=None):
"""
Args:
burn: The number of MCMC iterations to discard as burn-in.
Returns:
A pd.Series containing the marginal probability that each coefficient
is positive.
"""
if burn is None:
burn = R.suggest_burn(self.log_likelihood)
draws = self._coefficient_draws
probs = np.array(
[np.mean(draws[burn:, i] > 0) for i in range(draws.shape[1])])
return pd.Series(probs, index=self.xnames)
def inclusion_probs(self, burn=None):
"""
Args:
burn: The number of MCMC iterations to discard as burn-in.
Returns:
A pd.Series containing the marginal probability that each coefficient
is nonzero. The series is indexed by the set of variable names.
"""
if burn is None:
burn = R.suggest_burn(self.log_likelihood)
draws = self._coefficient_draws
probs = np.array(
[np.mean(draws[burn:, i] != 0) for i in range(draws.shape[1])])
return pd.Series(probs, index=self.xnames)
def suggest_burn(self):
return R.suggest_burn(-1 * self._residual_sd_draws)
