search

pymc-devs / pymc

Bayesian Modeling and Probabilistic Programming in Python
https://docs.pymc.io/
Other
8.49k stars 1.97k forks source link

Error in idata.posterior variable #7234

Closed logongas closed 3 months ago

logongas commented 3 months ago

Run the example in the line: idata.posterior["y_model"] = idata.posterior["Intercept"] + idata.posterior["x"] * xr.DataArray(x) show de error:

KeyError: "No variable named 'x'. Variables on the dataset include ['chain', 'draw', 'Intercept', 'slope', 'sigma', 'y_model']"

Description

change variable "x" by "slope"

Type of change

📚 Documentation preview 📚: https://pymc--7234.org.readthedocs.build/en/7234/

welcome[bot] commented 3 months ago

Thank You Banner] :sparkling_heart: Thanks for opening this pull request! :sparkling_heart: The PyMC community really appreciates your time and effort to contribute to the project. Please make sure you have read our Contributing Guidelines and filled in our pull request template to the best of your ability.

review-notebook-app[bot] commented 3 months ago

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

OriolAbril commented 3 months ago

Hi @logongas, thanks for the PR but sadly it is not needed. I'll try to explain why.

All the notebooks in https://www.pymc.io/projects/docs/en/stable/learn/core_notebooks/index.html are executed with every commit made to pymc to make sure they run with each change. All builds have been able to execute the notebook correctly.

What I think is happening to you is that you are using the initial pymc model, the one defined as:

with Model() as model:  # model specifications in PyMC are wrapped in a with-statement
    # Define priors
    sigma = HalfCauchy("sigma", beta=10)
    intercept = Normal("Intercept", 0, sigma=20)
    slope = Normal("slope", 0, sigma=20)

    # Define likelihood
    likelihood = Normal("y", mu=intercept + slope * x, sigma=sigma, observed=y)

    # Inference!
    # draw 3000 posterior samples using NUTS sampling
    idata = sample(3000)

in order to run the plots at the bottom. However, if you run the notebook sequentially you'll see that before getting to the plots, a roughly equivalent model is defined in bambi:

model = bmb.Model("y ~ x", data)
idata = model.fit(draws=3000)

These two lines do basically the same as the code cell above, but not all the variable names are the same in both models (even if meaning the same conceptually), nor all priors are the same (bambi sets defaults for them), nor the samples will be exactly the same either even if the observed data is the same because MCMC has a random component.

If you do want to use the first model you need to change how y_model is defined, as you have done, but the notebook executes the plots on the bambi model. Don't hesitate to let us know if you have any more doubts.

logongas commented 3 months ago

@OriolAbril Thank you very much for the reply. Your explanation is clear