Data Visualization Guide

Idioms are "a distinct approach to creating and manipulating visual representations” (Munzner 2014). In other words they are the form or type of visualization that you are choosing to represent the data. Common idioms include bar graphs, pie charts etc.

So once the data has been prepared, you are now ready to choose the best form for visualization. So are you going to use a scatter plot, bar graph, line graph or map? These are all considerations that you will have been thinking about in the first two stages, especially when you’re considering what type of data you have . If you’re having trouble still deciding, or don’t know what is the best form. There’s actually a really great tool called The Data Visualization Catalogue. In this tool you can click a button that says 'choose by function" and you can pick an idiom based on what type of relationship you want to show with your data.

So this is a great tool, but other questions that will help you is looking at how many dimensions your data has, and other types of specialized characteristics that your data contains. Borner and Polley identified 5 major categories of data in Visual Insights. Of course this isn't exhaustive and your data might side outside of this.

1. Temporal Data - This type of data answers the "when" question and highlights the temporal distribution of datasets; to identify growth rates, latency to peak times, or decay rates; to see patterns in time-series data, such as trends, seasonality, or bursts. 
2. Geospatial Data - This type of data answers the "where" question and uses location information to identify position or movement over geographic space.
3. Topical Data - This type of data is textual, linguistic or semantic data, often used in the humanities and social sciences. This data answers the "what" question.
4. Tree Data - Answers the question "with whom." Tree datasets, such as directory structures, organizational hierarchies, branch- ing processes, genealogies, or classification hierarchies are commonly organized and displayed using tree visualizations: for example, tree views, treemaps, or tree graphs
5. Network Data - Answers the question "with whom" as well. This type of data aims to increase our understanding of natural and manmade networks. This can look like social network analysis, bibliometrics, mapping in physics etc.

Going back to idioms,  sometimes you don’t need to choose the most complicated form. For example, you might want to start asking yourself – why not a bar graph? The reason why it’s so common is it’s a classic and it works, it’s quite easy to understand. It mixes both qualitative and quantitative data, easy to compare, easy to rank, easy to show deviations. One thing to consider though is it’s quite easy to manipulate data using bar graphs, you always want to make sure the axes start at zero, and consider whether the scale that you’re using is exaggerating the fluctuations.

Another common type is a scatter plot – what’s great about a scatter plot is that you fit a lot of points into a small space. For example if we showed this exact information in a bar graph it would’ve been way to much information to fit in without aggregating them. We use these because a it’s easy to give a sense of relationship, trends and also it’s easy to see outliers as well. If you want to add a third variable you can use something called a bubble chart: A Bubble Chart is a cross between a Scatterplot and a Proportional Area Chart. This combines three different dimensions to see overall patterns and compare the using relative size rather than scale.