There is no one size fits all in Tableau.
One of the foremost difficult things that I have found is to determine which chart suits you for the data you want to show. Tableau combines creative skill along with data to tell a story and it is up to you to find out what type of storytelling best suits your needs. The most effective kind of visualization is what answers the key questions in the most lucid way as possible.
· Who needs to look at the visualization? What are their key questions?
· Does the chart communicate the predominant concerns immediately and clearly?
· Is the chart too cluttered with data? Can the user isolate the data easily to highlight key information?
When you see a chart infographic about crime statistics for example, if the demographic representations in density and level are not immediately clear - the visualization fails to fulfill its main purpose and in a real-life scenario, user adoption of that visualization may fail.
We go back to choosing the chart type. The first step is to understand what chart type you can use.
That is often decided by what dimensions and measures are available to you.
Dimension is qualitative or categorical data. Measures is quantitative data which can be numeric, currency etc.
Tableau gives you the flexibility of interchanging datatypes but the scope of changing data types in the manner that is needed is an objective decision of the creator.
Another thing to consider is the green pill – blue pill feature of tableau.
Green pill is continuous data and blue pill denotes discrete data. Continuous data is treated as infinite and makes tableau understand that the data is continuously changing while discrete data has a finite range, and the consecutive values of discrete data are distinct and separate.
This is highly important, because more than dimension and measure which is logically more sensible to us, the discrete and continuous nature of the data helps tableau gauge what charts can automatically be plotted by Tableau.
One useful tool used for beginners is the ‘Show me’ feature in tableau where the chart type applicable becomes enabled based on the data selected. However, if one strictly works from the context of the 'Show Me' enabled feature, it may descope some important chart types available to use, if the data is not formatted or transformed correctly.
Take for example – a pie chart. Logically a discrete dimension and continuous measure should be enough for a pie chart, but as demonstrated below - the pie chart is not enabled in tableau even though the applicable criteria appears to be fulfilled. What makes it confusing is that logically this should work.
The reason this feature is disabled is because Tableau does not support the automatic creation for a pie chart for aggregate measures. Fortunately, there is an indirect way to do this.
Also, if the built-in chart types are applied, it may descope other chart types due to data manipulation coded into that chart type functionality. For example, the histogram feature automatically creates bins of the continuous measure and applies it against the count of the measure. Thus, the actual scope of charts using the ‘Show me’ is not correctly visible.
Observe below, the histogram is enabled as a ‘Show me’ feature.
Once the user clicks on the Histogram feature, the bar chart is no longer available as the context of data available has changed.
Now, we look at the supported chart type based on discrete dimensions and continuous measures. Please find a table here detailing the minimum and maximum criteria of some of the popular chart types of Tableau.
1. The ‘More’ values indicate that there may be ways to apply more level of detail to the chart and cannot be limited to a specific number.
2. For unavailable values, I could not determine a specific number for this criteria
After finding the charts available to you, if there are more than one, the one to choose is again what chart type is the most effective and how soon does the viewer come to the answer of their questions. It is tempting to go along with an unusual chart as it seems bar graphs answer most questions. This is where the creativity kicks in.
This is where the charts chosen can be evaluated based on the key questions and the proper chart may be selected on this. Furthermore, not all data needs to be represented in one chart and can be distributed among multiple charts to make the data visualization more effective and sensible.
Not all level of details need be captured on x and y axis or in dimensions or measures but in other very useful tableau features like detail, filters, highlights etc. The possibilities are endless and creativity using tools in tableau goes a long way towards effective data visualization.
Thus, to sum up, the key points to consider when choosing a chart type for your visualization are: -
Key questions to answer and who is the audience
What are the dimensions and measures (discrete and continuous)
What are the supported chart types
Breakdown of information into chart types if needed
What is the most effective way to represent data
What other additional features can contribute to the visualization