Good Charts
This tutorial introduces how to construct a good and meaningful ChartHow to Make Good Charts
The best practice of making any type of good chart is to give clear and concise context to the viewer, but what this means is at least partially subjective. Data visualization lacks in this area where many elements are inconsistent with the entire design. Design flaws with data visualizations include lacking meaningful captions and issues with visualizations being difficult to understand.
Simon et al. [1] suggest issues such as line width, alignment, lack of meaningful descriptions, etc. can all impact comprehension, meaning that they are not readily interpreted. Examining reports from ITiCSE and full papers, they noticed that data visualizations for computing education were difficult to understand due to unorganization and lack of context.
This means considering line width, number alignment, meaningful text, no [confusing] acronyms, and having a large font size contribute to understanding such charts. According to Simon et al. [2] a good rule of thumb when it comes to visualizing data is that for any diagrams, tables, and graphs, be sure to check whether they contain any text or numbers that are smaller than the text of the paper's copyright notice. If it is smaller, it is a good indication that the graph we made will not be readable to the viewer.
It is also worth noting that the information presented should also be meaningful, therefore it is not advised to over complicate data portrayals. It is better to have multiple simple graphs than one complex graph. The caption of charts should be to the point and give viewers an idea of what this data is without having to refer back to the text.
Another aspect to consider is the choice of colors when designing a chart. Color palettes should be meaningful as well as distinguishable from each other, especially if they are adjacent. As an example, if a scientist is recording average temperatures in various states in the U.S., such as warm colors to signify hot temperatures.
The following is a reference on when to use differing data representations [3]:
| Type of Chart | When should this chart be used? |
|---|---|
| Bar Graphs | To compare categorical data or summary statistics from 1+ groups |
| Line Graphs | To depict a single variable or multiple variables with respect to change over time |
| Histograms | To portray sampling distribution with continuous independent variables |
| Box Plot | To show the distribution of data of 1+ groups |
| Pie Chart | To show frequencies or percentages |
| Scatter Plot | To determine if two varibles have a relationship or correlation |
| Violin Plot | To visualize peaks and distibutions in numerical data |
References
[1]: Simple Simon, Brett A. Becker, Sally Hamouda, Robert McCartney, Kate Sanders, and Judy Sheard. 2019. Visual Portrayals of Data and Results at ITiCSE. In Proceedings of the 2019 ACM Conference on Innovation and Technology in Computer Science Education (ITiCSE '19). Association for Computing Machinery, New York, NY, USA, 51 57. DOI: https://doi.org/10.1145/3304221.3319742.
[2]: Franzblau LE, Chung KC. Graphs, tables, and figures in scientific publications: the good, the bad, and how not to be the latter. J Hand Surg Am. 2012 Mar;37(3):591-6. doi: 10.1016/j.jhsa.2011.12.041. Epub 2012 Feb 2. PMID: 22305731. http://people.eku.edu/ritchisong/tables_tigures.pdf.
Next Tutorial
In the next tutorial, we will discuss downloading CSVs, which describes how to download CSVs to use the data in the tutorials.