What is a good way to deal with defining multiple closely related PYMC models while avoiding duplicate code? For example , suppose I want to compare a model that uses Student_T versus Normal likelihood, and also I want to compare including certain features or not, etc. My current approach is to us…

I will often split out bits of my models into little functions for organization. A nice helper function is pm.modelcontext for this. You can make a function like this: def add_skew_likelihood(mu, sigma, y, alpha_mu=0, alpha_sigma=4, model=None): model = pm.modelcontext(model) with model: …

Dealing with multiple closely related models

Questions version agnostic

ricardoV94 February 25, 2025, 9:51pm 2

You may want different functions that build distinct parts of the model, but in the end it will look something like that yes

1 Like

Topic		Replies	Views
Composing models Questions	2	984	August 6, 2017
Building PyMC model programatically version agnostic modeling	4	589	November 23, 2022
Fitting multiple independent hierarchical models at once version agnostic modeling	2	510	August 5, 2022
Pass a Model defined in a config file to a later file Questions	2	483	September 9, 2019
Notation for hierarchical models Questions	5	1061	April 20, 2018

Dealing with multiple closely related models

Related topics