Hello, everyone. I was trying to run some models to perform a kind of reconciliation over multiple levels. The idea is to have a base level (a.k.a, lvl3) which is Poisson distributed with a given mean (lvl3). Later, this will be joined using the S matrix to generate the observed data of the followi…

with pm.Model() as model: l3 = pm.Poisson("l3", mu=lvl3, shape=lvl3.shape) l2 = pm.Deterministic("l2", S_2 @ l3) l1 = pm.Deterministic("l1", S_1 @ l2) l0 = pm.Deterministic("l0", S_0 @ l1) l2_pot = pm.Potential("l2_pot", pm.logp(pm.Poisson.dist(mu=lvl2, shape=lvl2.shape), l2)) …

Reconciliation via Poisson Sampling

Questions v5

Cristianobam December 21, 2023, 11:54am 6

How should I fix the problem with the second implementation? Going with the second implementation I can’t run…

Topic		Replies	Views
Bernoulli samples all the same? Questions	1	2449	May 29, 2018
Qualitatively different results depending on shape parameter Questions	1	424	October 21, 2017
Pm.sample_ppc breaks with AttributeError Questions	16	2027	February 5, 2018
Custom discrete distribution for zero truncated Poisson Questions	10	2608	December 19, 2017
Can I avoid sampling irrelevant bits of the distribution? Questions	1	481	February 2, 2020

Reconciliation via Poisson Sampling

Related topics