A lot has been written about writing Jekyll (the platform that generates this website – it’s frequently used with github pages), but I haven’t seen a one size fits all solution. This is how I’m doing it (thanks to Johannes Hellmuth for getting me started)

The magic happens in the yaml header of your rmarkdown file. Here’s the header from a recent post:

layout: post
title: "Splitting a Sample by Two Balancing Factors"
author: "Matthew Peverill"
date: "`r Sys.Date()`"
    variant: gfm
    preserve_yaml: true
knit: (function(inputFile, encoding) {
                      encoding = encoding,
                      output_file = file.path(paste0(
                                                  substr(basename(inputFile), 1, nchar(basename(inputFile)) - 4),

In addition to the standard fields, you need to have everything that your Jekyll website will expect (in my case layout: post). Then preserve_yaml makes it so that those yaml fields get passed to the md that is built by rmarkdown. The ‘knit: ‘ is a hook for the rstudio ‘knit’ button that changes the output file to save an md file in the correct location for my website, with todays date as a prefix (which is the naming convention I use for posts).

Then, in your setup block, include something like this:

knitr::opts_knit$set(base.dir = "~/Dropbox/mrpeverill-website/", base.url = "/")
knitr::opts_chunk$set(fig.path = "assets/img/BalancedSamplingTest/")

Obviously you will need to change the base.dir and fig.path as appropriate. base.dir needs to be absolute within your filesystem, fig.path needs to be relative to theh base.url. I like to store my images in a subfolder named after the post (because typically they just get named ‘unnamed chunk img 4-4’ or some such, which would conflict with other posts.