Chart type guidance

Below is a guideline on how to choose chart types.

Introduction

Data visualization is presenting large amounts of data graphically with the goal to uncover information directly to the audience. It’s not simply using visualization tools to turn data into graphs. Instead, it is looking at the world from a data point of view.

The most important reason for data visualization is to help people understand the data faster, and to draw their own conclusions at a glance. Using graphs, charts, or symbols can clearly communicate key insights. Good visualization should simplify messages and make the main points easy to understand.

Know what you want to say

Information can be visualized in a number of ways, so before designing any visualizations for your data set, ask these questions to figure out the sort of data you want to understand and why you might need a chart for the data at all?

What is the goal?
What patterns or insights is the data saying?
Is this for a presentation or a distributed (self-access) visualization?
What type of visualization should I use? *

*Different chart types work best for different types of information. Consider what the data is describing, what is the question the data answers, then choose a chart type that bests showcases that takeaway.

Principles

Please follow the below principles when creating a data visualizations.

Reference to Google’s Material DS.

Data Integrity (accuracy)

Prioritize data accuracy, clarity, and integrity, to presenting information in a way that doesn’t distort it.

Readability (helpful)

Help users navigate data with context and affordance that emphasize exploration and comparison.

Consistency (scalable)

Adapt visualization for different device sizes, while anticipating user needs on data depth, complexity, and modality.

Tips and tricks (General guidelines)

Aside from to always narrow down what the purpose of the visual is, here are some guidelines that would help your visualization better represent the data to your readers.

Use one color to represent each category
Order the dataset in a logical hierarchy
Use callouts to highlight important information
Aim to help readers compare values in the visual

Choose a chart type

The type of chart you choose depends mainly on two things:

1. The data you want to communicate

2. What point-of-view do you want to convey with the data

Comparing Values

Easily show the low and high values in a data set. Below charts are good for comparisons:

Bar

Column

Line

Scatter

Bullet

Change over time

Analyze trends over a period of time. Charts good for time series include:

Line

Bar

Timeline

Gantt

Sankey

Parts of a whole

Easily show individual parts that make up a whole. Charts good for proportions include:

Pie

Stacked bar

Area

Treemap

Ranking

Show an items position in an ordered list. Charts good for ranking include:

List

Bar

Column

Change over time

Analyze trends over a period of time. Charts good for time series include:

Line

Bar

Timeline

Gantt

Sankey

Distribution

Understand outliers, clusters, and range of values in a large dataset. Charts good for ranking include:

Box plot

Bar

Column

Line

Scatter/Bubbles

Flows

Show movements in the data between multiple states. Charts good for ranking include:

Sankey

Gantt

Chord

Network map

Interactive Behaviors

Dynamic charts provide interactive patterns that users can control over the data displayed.

Direct manipulation: Hide or show, expand, collapse, zoom, panning slides, pagination
Data controls: Filters, tabs, drop-downs, time frame selections
Empty States: Charts should display content when returns zero or error
display what to expect when the data is available
make clear indications that there is data missing

Page Layouts

Be it a dashboard view or presentation slides, a page’s purpose should reflect in it’s layout and interactive patterns. The design should reflect how it is used, whether it is for reporting high-level summaries or for deeper dive explorations of the data.

Prioritize the most important information up top (information hierarchy)
Display a focal point that points to special callouts (using color, position, size, and visual weight)
Arrange information based on the sequence of questions asked of the data.