Related to : Using WAIC to compute Log Predictive Accuracy Can someone please explain this behavior. I used NUTS to train linear regression model with a generated dataset that have 100 data instances and 100 dimensions. If we compute WAIC for the train data it reduces when increasing the number o…

Unexpected Results when using WAIC to Compute Log Predictive Accuracy

junpenglao May 13, 2018, 8:08am 4

You need to find out first what is causing the large var(logp) in WAIC or large khat in LOO. And depending on your model set up, there are different possibility that cause such problem. See for example:

Topic		Replies	Views
Using WAIC to compute Log Predictive Accuracy Questions	3	751	May 11, 2018
Calculating WAIC for models with multiple likelihood functions Questions	20	4626	October 26, 2020
Comparing ADVI and MCMC models using WAIC or LOO Questions	8	1948	April 3, 2018
Calculating model comparison using log pointwise predictive density Questions	15	3597	April 5, 2018
Is model with high or low WAIC and LOO better version agnostic arviz	1	1638	August 12, 2021

Unexpected Results when using WAIC to Compute Log Predictive Accuracy

Related topics