How to Use Heat Map Charts to Recognize Patterns in Complex Data

Analyzing and making sense of large volumes of raw data can make anyone’s head spin. This is why it’s important to take a step back and get a generalized view of the data before diving into more granular analysis. A great tool for distilling insights from massive datasets – and one that’s particularly handy for visual comprehension – is the Heat Map.

For those of you familiar with business intelligence (BI) and data visualization, if I were to mention the term ‘Heat Map’, each one of you may envision something entirely different. Despite your daily exposure to a dizzying array of charts and graphs, and your knowledge of BI terminology, it’s still highly plausible multiple Heat Map variants would come to mind.

And that’s completely OK! What many people don’t realize, is that Heat Maps are not data visualizations in their own right, per se, but are visualization properties that can actually be applied to a wide variety of data visualizations to better illuminate trends, outliers and patterns in your data. They also happen to be highly effective at doing their job. One of the more ignored facets of data visualization, and data-discovery for that matter, is in fact pattern recognition, and Heat Maps bridge that gap marvelously.

While this blog will focus specifically on what we at Dundas refer to as a ‘Heat Map Chart’ (emphasis on Chart), there are at least three data visualizations that pair well with Heat Maps that I can think of off the top of my head.

 

What is a Heat Map Chart?

Let’s get down to business. If you’re still reading this, it’s because you have a great deal of complex data and are intrigued by how a Heat Map Chart can be leveraged to reveal patterns within it. That, or you’re looking to add some heat to your map visualizations (if you’re interested in that type of Heat Map, don’t sweat it, you’ll definitely want to read this blog to ensure your map visualizations never fizzle out).

So, what exactly is a Heat Map Chart? Well, it’s essentially a tabular matrix that leverages color variety and intensity to visualize and examine complex, multi-variate data. By displaying your variables both horizontally and vertically and coloring the newly created cells according to some pre-defined logic, you’re able to better see variance, patterns, and correlations across the data set (if any exist).

Now, rather than relying on my descriptive capabilities (or lack thereof), let’s take a look at an example of a Heat Map Chart that’s been built using Dundas BI to understand what one could look like:

Wow – beautiful isn’t it! In a typical Heat Map Chart – as is the case in this example – each axis will be used to display one category (for example, Day of Month along the bottom, and Hour of Day along the side). The individual rows and columns will then be further divided into color-coded subcategories (or cells) based on the value they contain, which is based on their relationship between the aforementioned intersecting rows and columns.

 

How is a Heat Map Chart Read?

When analyzing numerical data using this type of Chart, the differences between low and high values are noticeable due to the varying intensities of color. The more intense the color in a cell, the higher the value. This, in essence, is how a Heat Map Chart should be read.

What you may notice, is that the Color Scale (a legend I cannot recommend enough for this chart type) accompanying the Heat Map Chart uses a gradient scale from 0 to 1400 (we’ve gone ahead and broken it down visually in increments of 200 for the viewer to better grasp the change in values).

Remember earlier when I mentioned Heat Map Charts are great at generalizing large volumes of data? No? Paragraph 1, line 3! Well, now you can start to understand why. Seeing as they rely heavily on color to communicate values, it can be hard to accurately identify the differences between shades of color and isolate individual data points. This isn’t to say that the Heat Map Chart can’t be used for more granular analysis, it’s just better suited for high-level generalizations and can be used as a springboard for further inquiries.

 

Build a Heat Map Chart in Dundas BI

That covers the what and how of Heat Map Charts; if you’re interested in building your own, take a look at this video from Dundas’ Senior Solutions Architect, Jeff Hainsworth.

In this great video, we cover how to read, understand and build a Heat Map Chart, and go over some unique examples that were built using Dundas BI.

About the Author

Jordan Zenko

Jordan Zenko is the Community & Content Manager at Dundas Data Visualization. As Dundas’ resident (and self-proclaimed) story-teller, he authors in-depth content that educates developers, analysts, and business users on the benefits of business intelligence.

Follow on Linkedin
Print