In many of the datasets with which I am working, I have observed that the mode of the parameter posterior often differs significantly from the MAP estimate. I am curious to know the common reasons for this phenomenon and what factors I should investigate. I assume a reasonable answer would be as fo…

Note that the MAP is the mode of a the full, multivariate posterior distribution. In your case, this posterior lives in a parameter space that is at least 13-dimensional. It is not possible to visualize this posterior distribution. Instead, we visualize low dimensional “projections” of this high d…

Common reasons for getting a map estimate that is far from the mode of the posterior

Questions v5

hughlaughter April 7, 2023, 4:32am 3

edited the question, check “seasonalityday-0-32-google” and “saturation-0-32-google” in the plot above.

Topic		Replies	Views
How to reference posterior mode value without .find_MAP? Questions	3	1247	August 8, 2019
Getting point estimate from posterior Questions	2	1761	January 22, 2021
Simple exponential model fails to find map_estimate Questions	14	630	July 2, 2021
Posterior predictive sampling with data variance Questions	10	2423	September 14, 2018
Model selection for models with time-consuming sampling and large datasets version agnostic	0	423	February 2, 2023

Common reasons for getting a map estimate that is far from the mode of the posterior

Related topics