Data Storytelling — How to Choose the Right Chart or Graph for Your Data
Reference : help.tableau.com
What chart or graph works best for your data? In Tableau, form follows function. The visualization (or viz) you create depends on:
- The questions you are trying to ask
- The properties of your data
- How you want to present and communicate your insights to others
For example, showing the growth in sales each year requires a different visualization than showing the connection between discounted items and their profitability. Knowing what you need to show will help determine how you want to show it.
This topic presents nine different types of information that you can display with a visualization. This isn’t a comprehensive list, and there are bound to be exceptions to these categories. With experience you will be able to more quickly assess what chart type you want to create. Because Tableau is flexible, we encourage you to think outside the box. However, before you think outside the box, it’s helpful to start with some common chart types.
Change over time
Showing a change over time for a measure is one of the fundamental categories of visualizations. There are many options for exploring change over time, including line charts, slope charts, and highlight tables.
To show change over time, you need to know the value you expect to change, and how to work with Date fields in Tableau.
What kind of question does this chart answer?
- How has this measure changed in the past year?
- When did this measure change?
- How quickly has this measure changed?
Correlation
Sometimes you have two variables and are looking for the relationship between them. For example, you may be looking for the relationship between classroom size and school graduation rate, or how much lung capacity relates to endurance. (But remember, correlation does not always equal causation.)
Correlation can be shown with scatter plots or highlight tables, and you can use Tableau’s analytics objects(Link opens in a new window) to show the strength of the correlation.
What types of question can this chart answer?
- Are these two measures related? How strongly?
- Are some measures more related than others?
- How strongly related are these measures?
Magnitude
Magnitude shows the relative size or value of two or more discrete items. If you are comparing sales for different regions, you are looking at magnitude.
Magnitude charts include bar charts, packed bubble charts, and line charts.
What types of question can this chart answer?
- Which of these dimension members has the highest measure?
- Are there any exceptional dimensions?
- How large of a gap is there between the lowest and highest measure between these dimensions?
Deviation
Deviation charts show how far a value varies from some baseline, such as the average or median. If you wanted to know which items had unusually high or low profit margins, you would use a deviation chart.
You can use bullet charts, bar charts, and combination charts to show deviation. You can also find the statistical significance of the deviation using a Z-score.
What types of question can this chart answer?
- How far from the norm does this measure stray?
- How important are the deviations in this measure?
- Is there a pattern to the deviations?
Distribution
When you are trying to find the frequency of events within a population, you are looking at the distribution. If you are showing the number of respondents to a survey by age, or the frequency of incoming calls by day, a distribution chart might be the best choice.
Distribution charts include histograms, population pyramids, Pareto charts, and box plots.
What types of question can this chart answer?
- Are events clustered around a certain probability?
- Which population group buys the most items?
- When are the busiest times in our work day?
Ranking
Sometimes you not only want to depict the magnitude of some value, but also the relative ranking of all the members of your dimension. Showing the top ten sales people or demonstrating the under-performing states use a ranking chart.
Ranking charts are usually bar charts that integrate rank calculations, top n sets, or key progress indicators.
What types of question can this chart answer?
- How many people are under-performing in the company?
- How much revenue is generated by our top ten customers?
- What is the value of our ten lowest revenue properties?
Part-to-Whole
Part-to-Whole charts show how much of a whole an individual part takes up. For example, if you are showing how much each region contributes to overall sales, or how expensive each different shipping mode is for an individual product, you would use a part to whole chart.
Part-to-Whole charts can be pie charts, area charts, stacked bar charts, or treemaps.
What types of question can this chart answer?
- How much does this value contribute to the total?
- How does the distribution of costs change each year?
- Do different items contribute different amounts to sales by region?
Spatial
Spatial charts can precise locations and geographical patterns in your data. Showing the airport terminals with the most foot traffic or a map of all sales across the country are examples of spatial maps.
Spatial maps include filled maps, point distribution maps, symbol maps, and density maps.
What types of question can this chart answer?
- Which city has the highest sales?
- How far from distribution centers are our customers?
- How many people arrive at which gate?
Flow
Flow charts can be maps that convey movement over time, such as Sankey diagrams. Flow maps include path over time and path between origin and destination charts.
What types of question can this chart answer?
- What is the longest shipping route?
- How long are people lingering around gates?
- What are the bottlenecks to traffic in the city?