# Presenting data of complex tables

I finished some experiments on classification and I am now preparing for a presentation.

The experiment yields data which will then be manually verified to determine if the data is correct or incorrect. The experiment is run a total of 10 times and the output for each run is the amount of hits ("Yes") or misses ("No"). In the end, I have this complex table:

• What is the best way to present this kind of complex tables?
• If applicable (as in my example), should I present normalized data?
• If applicable (as in my example), should I present average values?
• Depends how much information you want to display. Boxplots can also be an option. – Marc Claesen Dec 3 '13 at 20:01

I believe visual representation is nearly always a much better way to present data-intensive information. However, it may depend on your audience, and the depth of analysis you go into with your results. If you are repeating each experiment n times, it would be most useful, in the least, to present the average result (the mean), along with a standard deviation. There are many different ways to visually represent your data, but whatever method you use must be able to deliver the main points you need to convey clearly. To help you decide on the best form of presentation of data, here a couple of suggestions:

• Ask yourself what the main message is that you need to deliver from this data, and focus on a delivery style (graph, table, list, whatever) that best delivers that main message (and not all the other stuff)
• Reduce the number of variables you need to display to reduce distractions from that main message. So you wouldn't need to display results from 10 repeats of an experiment when you can just show the average result for that experiment.
• Remember that in the majority of cases, a picture tells a thousand words.

As an aside, regarding your statement "The experiment yields data which will then be manually verified to determine if the data is correct or incorrect." Always believe your data. Your data is always correct. You just have to come up with the explanations for why the data is as it is.

If you are going to show frequency data as yours with more than 4x4 cells in a presentation, I would generally recommend to use graphs instead of tables. Either you spend much time explaining such a big table or you will leave your audience with the feeling that they just saw something they did not have time to really understand.

The LabWrite Guide by NC State and the paper "Using Graphs Instead of Tables in Political Science" by Jonathan Kastellec and Eduardo Leoni provide some ideas for graphs substituting tables.

Note that there is a closely related question about Book recommendations for information visualization?.