The problem with discrete approximation to integrals or probability densities @SpartacanUsuals

Uploaded May 2018 | Updated September 2026, 2 hours ago

One solution to approximating high dimensional integrals involves discretising. These approaches involve approximating a continuous function as a discrete function, with a finite set of values at particular points. This video explains the issue inherent with this approach to doing approximate Bayesian inference.

This video is part of a lecture course which closely follows the material covered in the book, "A Student's Guide to Bayesian Statistics", published by Sage, which is available to order on Amazon here: amazon.co.uk/Students-Guide-Bayesian-Statistics/dp/1473916364

For more information on all things Bayesian, have a look at: ben-lambert.com/bayesian/. The playlist for the lecture course is here: youtube.com/playlist?list=PLwJRxp3blEvZ8AKMXOy0fc0cqT61GsKCG&disable_polymer=true

The problem with discrete approximation to integrals or probability densities

How to derive a Gibbs sampling routine in general

Derivation of variance-covariance matrix in factor analysis - part 2

Factor analysis: predicted variance and covariance of indicators - part 1

Monte Carlo Simulation of Omitted Variable Bias in Least Squares

Monte Carlo Simulation for Ordinary Least Squares

How to code up a bespoke probability density in Stan

Maximum likelihood estimation of factor analysis models - part 2

Why we typically use dependent sampling to sample from the posterior

An introduction to the Poisson distribution - 2

Score test (Lagrange Multiplier test) - introduction

Representing heteroscedasticity in matrix form