Jake Graving
@jgraving.bsky.social
Interested in behavior and computation. Research Scientist at Max Planck Institute of Animal Behavior
If you mean weighted log liklehood should be able to use logp and Potential or alternatively CustomDist. Something like:
rv = pm.Distribution.dist(params)
ll = pm.logp(rv, y)
pm.Potential("w_ll", (ll * w).sum())
www.pymc.io/projects/exa...
rv = pm.Distribution.dist(params)
ll = pm.logp(rv, y)
pm.Potential("w_ll", (ll * w).sum())
www.pymc.io/projects/exa...
www.pymc.io
September 12, 2025 at 9:56 PM
If you mean weighted log liklehood should be able to use logp and Potential or alternatively CustomDist. Something like:
rv = pm.Distribution.dist(params)
ll = pm.logp(rv, y)
pm.Potential("w_ll", (ll * w).sum())
www.pymc.io/projects/exa...
rv = pm.Distribution.dist(params)
ll = pm.logp(rv, y)
pm.Potential("w_ll", (ll * w).sum())
www.pymc.io/projects/exa...