How to chart?
We humans are visual creatures by nature.
MIT research shows that the human brain processes images in 13 to 33 milliseconds[¹], which is 60,000 times quicker than text[²]. The brain also favors visuals, as they make up to 90% of the data it receives[¹].
Visuals generate far more trust.
While mere words or numbers persuade only about 68% of people, adding a simple chart boosts the trust level to an astounding 97%[¹].
When executed correctly, visualizations are powerful tools.
They cut through the noise, revealing the hidden story in the data.
Charts distill large amounts of data in a digestible format, making them ideal for sharing data with stories with non-technical audiences.
If you’ve ever felt swamped staring at a massive spreadsheet, you’ll know how helpful visuals can be.
How can we create charts that are engaging, relevant, and well-received by diverse audiences?
Effective data visualization requires balancing aesthetics and functionality.
Blending analytical depth with compelling storytelling demonstrates a knack for visual storytelling that goes beyond adding superficial embellishments.
The data and its visual representation must complement each other.
To do this, you must be crystal clear on the takeaways you want to convey.
Your visuals should have a defined purpose, and you should include only the elements required to achieve it and leave out anything extra.
Start from your audience:
- What do they care about?
- What do they need to know?
- What questions and concerns they might have?
Then turn to your analysis:
- What interesting insights can you deliver?
- What helpful discussions could your visuals spark?
- How can you anticipate their questions and concerns?
Matching visuals to analytical goals
Choosing the right visual requires clarity on your analytical goals:
- Are you examining how parts contribute to a larger whole?
- Exploring relationships between variables?
- Exploring trends over time? Or,
- Comparing and ranking data?
In the next sections, we’ll pair up each question with the visuals that best illustrate them.
And if you’re keen to start charting, take a look at this Data Visualization Tutorial in Hyperquery.
This data analytics platform wraps the powerful Vega-lite[³] charting language in a dead-simple UX.
This makes beautiful, pro-level charts totally accessible - no data viz PhD required! 🤓
Tracking trends over time
Line, area, and bar charts are the go-to choices for visualizing trends over time.
- Line charts depict continuous trends,
- Bar charts are ideal for comparing categorical data,
- Area charts display cumulative trends while emphasizing magnitudes,
Consider this example tracking music sales from 1973 to 2019.
The x-axis represents the years, while the y-axis represents sales amount.
Notice the bold blue line in 1999 depicting the peak of physical sales, with the ascending purple line in the early 2000s showcasing the emergence of digital streaming.
This visualization effectively captures the trends of music sales over time and the relative performance of different formats, but how do these sales figures compare?
Comparison and ranking
We rank data all the time.
Sales reps' productivity, segments’ contribution, channels’ profitability.
Bar charts are perfect to visualize rankings.
They place values along a shared baseline, making it really easy to compare items side-by-side.
The chart below answers the question of total sales and total sales by format at a glance.
Area charts also compare values across categories.
But there’s a key difference with bar charts.
The area chart displays each format as a single color-coded pattern across the entire time range. This emphasizes the trajectory of each format.
In contrast, the bar chart depicts each year as a segmented bar broken down by format. This emphasizes year-to-year comparison across each format.
Relationship between variables
- Does app usage correlate with user loyalty?
- Does website loading speed correlate with conversion rates?
- Does the number of product features correlate with customer satisfaction?
Analytics is a perpetual quest trying to uncover meaningful patterns in the data.
Scatterplots are an ideal starting point.
The chart below shows a positive correlation between songs’ energy and danceability.
The cluster in the upper-right corner reveals songs with both high energy and high danceability.
However, there is also a lot of variation in the data.
High-energy songs coexist with low danceability, and there are low-energy compositions surprisingly danceable.
This hints that danceability isn’t solely determined by energy levels.
For the music nerds out there, we all know that other factors like tempo, rhythm, and melody are part of the equation.
Heatmaps use color-coding to spotlight trends and outliers within large datasets at a glance.
Take this chart as an example.
It vividly illustrates that the most prevalent tempo range is 160+ BPM, closely followed by 140–160 BPM.
When it comes to energy, the 0.8–1.0 range takes the top spot, with 0.6–0.8 next.
Nost songs trend fast-paced and moderate-high energy dominates.
Part-to-whole relationships
Can you tell which artist got the most views?
Here comes the hot take you’ve been waiting for 🚨
use pie charts sparingly and only as supporting elements.
The issue is our visual system struggles with surface areas, particularly when comparing non-adjacent slice sizes to each other.
This limitation makes pie charts a useful supplement when paired with a more intuitive chart.
But on their own, they lack impact.
Instead, a bar chart immediately directs the eye to the most viewed artists.
Stacked bar charts are great for portraying percentages.
In this graph, the evolution of music sales performance by format is instantly clear.
Don’t hesitate to lean heavily on bar charts.
Their simplicity and versatility facilitate meaningful comparison across virtually any dataset.
Conclusion
Poorly visualized data leads astray.
Yet poor visuals abound, wasting the resources invested in collecting and processing the data.
When done right, data visualization distills nuanced findings into accessible insights anyone can grasp.
An invaluable skill in today’s data teams.
In this post, we walked through a simple 3-step process to chart creation:
- Start with your audience in mind,
- Define your analytical goal,
- Match the visual to the goal.
Follow these steps, and your visualizations will speak clearly.
Visualization offers a chance to enlighten and inspire.
The power is yours — use it wisely.
Happy visualizing 🎨🤓
— -
[¹]: [In the blink of an eye | MIT News | Massachusetts Institute of Technology](https://news.mit.edu/2014/in-the-blink-of-an-eye-0116)
[²]: [MIS Research Center | Carlson School of Management (umn.edu)](https://carlsonschool.umn.edu/faculty-research/mis-research-center)
[³]: Vega-Lite is a grammar-based visualization library similar to ggplot and Altair.
