My PhD Project

Introduction

I-priors (Bergsma, 2019) are a class of objective priors which make use of the Fisher information. Estimation is simple, inference straightforward, and often gives better predictions for new data.

Regression

Simple fitting of various regression models for prediction and inference.

Classification

Extension to categorical responses for binary and multiclass classification.

Variable Selection

A fully Bayesian approach to variable selection for linear models.