models.BayesGlmFit
A sampled Bayesian GLM, from BayesGlm.fit.
Usage
models.BayesGlmFit()Attributes
| Name | Description |
|---|---|
| chains | Number of chains. |
| coefficient_draws | Coefficient draws, one row per draw (chain by chain). |
| dispersion_draws | Dispersion draws, one per draw (constant when fixed). |
| divergences | Divergent transitions among the kept draws. |
| names | Coefficient names. |
| posterior_mean | Posterior means of the coefficients. |
chains
Number of chains.
chains: int
coefficient_draws
Coefficient draws, one row per draw (chain by chain).
coefficient_draws: list[list[float]]
dispersion_draws
Dispersion draws, one per draw (constant when fixed).
dispersion_draws: list[float]
divergences
Divergent transitions among the kept draws.
divergences: int
names
Coefficient names.
names: list[str]
posterior_mean
Posterior means of the coefficients.
posterior_mean: list[float]
Methods
| Name | Description |
|---|---|
| log_likelihood() |
Pointwise log-likelihood of y given design: one row per draw,
|
| loo() |
PSIS-LOO of y given design, with each observation’s relative
|
| predict() | Posterior mean of each row’s mean. |
| predict_distribution() |
Posterior predictive draws across the rows, keyed row = 0, 1, ....
|
| summary() | Posterior summary: one dict per parameter with name, mean, |
log_likelihood()
Pointwise log-likelihood of y given design: one row per draw,
Usage
log_likelihood(design, y)Returns
list of list of float
loo()
PSIS-LOO of y given design, with each observation’s relative
Usage
loo(design, y)efficiency estimated from the chains.
Returns
Elpd
predict()
Posterior mean of each row’s mean.
Usage
predict(design)Returns
list of float
predict_distribution()
Posterior predictive draws across the rows, keyed row = 0, 1, ....
Usage
predict_distribution(design, n_sims, seed)Returns
PredictiveDistribution
summary()
Usage
summary()sd, q05, q50, q95, rhat, ess_bulk and ess_tail; the dispersion last when it was sampled.
Returns
list of dict