I have a simple inference example for multiple coinflips for multiple agents. For reproducibility I’m using p=11% for the probability of heads for all agents: import pymc as pm import arviz as az import numpy as np num_trials = 500 num_agents = 60 true_p_heads = np.array([0.11]*num_agents) observ…

A sum over n terms doesn’t need to be at all proportional to n. Take for example \sum_{i=1}^{n} \frac{1}{n} p_waic is in fact similar to this, for simple models it can be interpreted as the number of parameters, so it is also basically constant with n. In more complex models like hierarchical on…

P_waic of az.waic does not change proportionally with number of trials

Questions v5

ricardoV94 March 30, 2024, 10:15am 2

Maybe @OriolAbril can help here

Topic		Replies	Views
Calculating WAIC in hierarchichal models Questions	5	1380	June 9, 2020
WAIC values differing between identical fits? Questions	0	520	July 22, 2019
Unexpected Results when using WAIC to Compute Log Predictive Accuracy Questions	3	688	May 13, 2018
WAIC fails with AttributeError Questions	7	1591	January 10, 2018
WAIC and In-Sample Deviance Questions	4	1293	October 3, 2018

P_waic of az.waic does not change proportionally with number of trials

Related topics