Historian and scholars alike can represent their primary sources into data in many different ways. Based on my accrued knowledge I would focus you to represent and chose to go a more direct and linear approach to your data representation. As opposed to sticking to an almost entirely numerical data set. For example a university class could be used as a good example. We could in turn compile a linear dataset, a timeline with various data points and columns being shown over the hypothetical semester. An for potential vertical column categories could be “learned information vs information retained over time” and another could be data absorbed by different primary host of the particular lectures and so on for diff columns. Until you have enough data points to satisfy your requirements for the compilation. Which bring me into topic of visuals, and timeline is a great example of charts or graphs that could be used. But would be hard to adapt to a computer database. A what we view as a easy task and what is hard not the same for a computer. For instance what order your columns to flow may be hard for a human eyes to pick up but be no problem for a scanner or mass compiler to absorb your data into whole databases and networks.
One of the main advantages to me of using primary source as opposed to other sources is the ability to stand on it on legs. What I mean by that is that as a researcher in charge of your data and how your compiling that information to the public. You do have a duty to validate that said source material, but when using direct verbiage or other accounts taken directly from primary sources. The in liability in a way, or redirect it way from you as a research and allow that source to “ Stand on its on merit”. Also when you are utilizing primary source you are not absorbing that role as a interpreter and in a way you more or less remain a neutral academic. So by sticking with primary source data you are not filtering or silencing the past. But instead allowing the source material to speak for themselves and allow the audience to talk away what they will.
Wickham’s 3 Principles of tidy data were: Data Structure (footprint and layout) , Data Semantics , and Tidy Data (organizations and flow).