Skip to content

update to use pm.Truncated #423

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Sep 22, 2022
Merged

update to use pm.Truncated #423

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
250 changes: 147 additions & 103 deletions examples/generalized_linear_models/GLM-truncated-censored-regression.ipynb
View file
Open in desktop

Large diffs are not rendered by default.

24 changes: 9 additions & 15 deletions myst_nbs/generalized_linear_models/GLM-truncated-censored-regression.myst.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,7 @@ kernelspec:
(GLM-truncated-censored-regression)=
# Bayesian regression with truncated or censored data

:::{post} January, 2022
:::{post} September, 2022
:tags: censored, censoring, generalized linear model, regression, truncated, truncation
:category: beginner
:author: Benjamin T. Vincent
Expand Down Expand Up @@ -126,7 +126,7 @@ def linear_regression(x, y):
with pm.Model() as model:
slope = pm.Normal("slope", mu=0, sigma=1)
intercept = pm.Normal("intercept", mu=0, sigma=1)
σ = pm.HalfNormal("σ", sd=1)
σ = pm.HalfNormal("σ", sigma=1)
pm.Normal("obs", mu=slope * x + intercept, sigma=σ, observed=y)

return model
Expand Down Expand Up @@ -163,7 +163,7 @@ az.plot_posterior(trunc_linear_fit, var_names=["slope"], ref_val=slope, ax=ax[0]
ax[0].set(title="Linear regression\n(truncated data)", xlabel="slope")

az.plot_posterior(cens_linear_fit, var_names=["slope"], ref_val=slope, ax=ax[1])
ax[1].set(title="Linear regression\n(censored data)", xlabel="slope")
ax[1].set(title="Linear regression\n(censored data)", xlabel="slope");
```

To appreciate the extent of the problem (for this dataset) we can visualise the posterior predictive fits alongside the data.
Expand Down Expand Up @@ -217,15 +217,8 @@ def truncated_regression(x, y, bounds):
slope = pm.Normal("slope", mu=0, sigma=1)
intercept = pm.Normal("intercept", mu=0, sigma=1)
σ = pm.HalfNormal("σ", sigma=1)

pm.TruncatedNormal(
"obs",
mu=slope * x + intercept,
sigma=σ,
observed=y,
lower=bounds[0],
upper=bounds[1],
)
normal_dist = pm.Normal.dist(mu=slope * x + intercept, sigma=σ)
pm.Truncated("obs", normal_dist, lower=bounds[0], upper=bounds[1], observed=y)
return model
```

Expand Down Expand Up @@ -260,7 +253,7 @@ def censored_regression(x, y, bounds):
with pm.Model() as model:
slope = pm.Normal("slope", mu=0, sigma=1)
intercept = pm.Normal("intercept", mu=0, sigma=1)
σ = pm.HalfNormal("σ", sd=1)
σ = pm.HalfNormal("σ", sigma=1)
y_latent = pm.Normal.dist(mu=slope * x + intercept, sigma=σ)
obs = pm.Censored("obs", y_latent, lower=bounds[0], upper=bounds[1], observed=y)

Expand All @@ -284,8 +277,8 @@ with pm.Model() as m:
with pm.Model() as m_censored:
pm.Censored("y", pm.Normal.dist(0, 2), lower=-1.0, upper=None)

logp_fn = m.logp
logp_censored_fn = m_censored.logp
logp_fn = m.compile_logp()
logp_censored_fn = m_censored.compile_logp()

xi = np.hstack((np.linspace(-6, -1.01), [-1.0], np.linspace(-0.99, 6)))

Expand Down Expand Up @@ -368,6 +361,7 @@ When looking into this topic, I found that most of the material out there focuse
## Authors
* Authored by [Benjamin T. Vincent](https://github.com/drbenvincent) in May 2021
* Updated by [Benjamin T. Vincent](https://github.com/drbenvincent) in January 2022
* Updated by [Benjamin T. Vincent](https://github.com/drbenvincent) in September 2022

+++

Expand Down