A sample size based comparison of the frequentist and Bayesian logistic regression

Tissing, Arend-Jan (2021) A sample size based comparison of the frequentist and Bayesian logistic regression. Bachelor's Thesis, Mathematics.

Preview	Text bMATH_2021_TissingAJ.pdf Download (746kB) | Preview
	Text toestemming.pdf Restricted to Registered users only Download (120kB)

Abstract

This research studies the performance of the frequentist and Bayesian logistic regression. When comparing the performance of both methods, the effect of the sample size is discussed, to see whether there is an effect in the difference in performance when the sample size is decreased. The measure of performance used in this research is the Area Under the Receiver Operating Characteristic curve, which measures a model's ability to classify data with a binary outcome variable. 5-fold cross-validation is used when doing this comparison of different methods. Before doing the sample size reduction, within each of the approaches, multiple methods are compared. For the frequentist approach, these are the AIC, BIC and p-values methods and for the Bayesian approach, these are the MCMC and RJMCMC algorithms. After comparing these, to study the performance when decreasing the sample size, the data set is split in half multiple times and both regression approaches are performed on each individual part. Throughout this research, the Framingham data set is used as a benchmark data set.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Grzegorczyk, M.A. and Krijnen, W.P.
Degree programme:	Mathematics
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	20 Jul 2021 13:23
Last Modified:	20 Jul 2021 13:23
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/25356

Actions (login required)

View Item