Uncovering the heatmap history

Heatmaps are unavoidable nowadays if we want to speak or write something about web analytics. Numerous sites on the internet provide heatmap tools for other businesses who want a 2 or 3D representation of qualitative data. Mostly these are about their users’ habits, to filter the mistakes and make things better. 

Although other tools must do the math first, various data are well-represented through heatmaps. Moreover, a non-expert can also easily understand the exact way this tool works, also graphical representations can be both eye-catching and interesting, so the user recognizes them better later on. 

The name of the heatmap refers to the color which is the most essential part of the tool. In our mind, heat is associated with the red color since people first met with fires and flames. The expression “heatmap” tries to make a benefit from this association. In this graphical tool, the “warm” or red parts are usually the most interesting elements. Those are the busiest, highest or most influential parts of the map, it depends on how the creator imagines it and what type of heatmap he uses. Thus, there are almost infinite types of heatmaps; the limitations are our imaginations. People created heatmaps to understand sports better, to examine biological correlations, to represent weather reports, or, like I said, for web analytical reasons. 

You can read many articles and guides about the usage of heatmaps as every corner of the internet is full of them. However, somehow fewer people mention its history and impact. Too bad, because it hides a tremendous amount of stories which are both exciting and interesting. It will fly us back to the 19th and 20th centuries, when there was no internet and websites. How did they yet develop this whole graphical visualization and shape it for 150 years long? In this article, you find all the answers!

The first few steps

Like I said before, if we want to understand everything about how people invented heatmaps, we have to travel back to the late 19th-century of Europe.

Basically, there is just one way to explore the history of a tool: break it down into its elements and see where and how it started to become one. The first part of the heatmap, also some of us may say the heart of the tool, is the color-shaded matrix display. It’s like the wheels of the car: without them, you won’t use the car itself, since it can’t carry you from one place to another. The situation is quite similar here: without the colors, the heatmap doesn’t make sense.  

The first famous color-shaded matrix display belongs to Toussaint Loua, who was a well-known French statistician and tried to analyze social statistics (for example: national origin, profession and age) in Paris. He used 20 different maps to fit in one and make the whole implementation so unique. He drew this matrix with his own hands, also there were four main colors that he used: white to yellow and blue to red; and of course, the mixtures of the four colors. This whole story happened in 1873, so it’s easy to say that this one was the first color-shaded matrix in the world, right? Probably. But accountants and mathematicians had used a similar, but simplified method since the start of the 19th century usually to indicate the areas of interest. Unfortunately, they didn’t write their name down, thus nobody could identify them. 

Loua’s first heatmap 

Source

At the start of the 20th century, people started to both use Lola’s technique and complete with other elements, like permuting matrices. Permuting matrices means that in one row and one column, you must only place one specified value, but that value can belong to all the characteristics that the given columns and rows mean. It’s great for creating easy and understandable illustrations not as Lola did. 

Make it simpler and simpler!

40 years later, Willard C. Brinton was one of the most skillful professors, who had the ability to demonstrate diagrams understandable by everybody who was capable of reading. He made a book called Graphic methods for presenting facts in 1914, where he demonstrated his whole idea in a few words: “It is most desired to reach are those who have never had any statistical training, consistent effort has been made to keep the whole book on such a plane that it may be found readable and useful by anyone dealing with the complex facts of business or government.”

Two pages of Williar C. Brinton’s famous book

Source

The first half of the 20. century spent with a lot of experimentations, what are the easiest ways to present matrices and also how to simplify it, to be readable for any human being. 

The whole formula was as easy as usual: the more people attended elementary and secondary schools, the more people had the ability to know the meaning of these illustrations. Luckily, the huge improvements in educational systems all over the world helped the spread of these illustrations and diagrams. 

3. The relation between The Guttman scalogram and the heatmap

The next well-known exploration is not really related to the world of heatmaps, although it influenced many statisticians who later found out how the two different tools connect to each other, so we shouldn’t miss it! The Guttman scalogram was one of the most influential parts of 20th-century statistics analysis. Just imagine a bunch of questions which are related to each other, but we don’t know the exact order. But we are sure that one comes from another. It is a type of ordinal scale, which means that the responses are ranked in a specific order.

For example:

• I am not afraid of the dark 
• I love going out at nights
• I usually love to go for a walk alone in the evenings
• I wouldn’t be afraid if I got lost in a forest in the dark.

Then ask several “judges” who will rate the statements. They either can give a “1” (Yes) value if the statement is favorable to them or a “0” (No) if it’s not. Then we create a matrix from their answers and add up their results. 

Each judge gets a rank, the more “1” value they answer the higher they go on the ladder. Then we examine the questions. The more “1” value a question earns, the earlier it appears in the matrix. Our hypothesis in this case is that if one judge didn’t give ‘Yes’ value to the second liked statement, he dismissed the third liked one and so on. It is usually true, but of course, there are exceptions. 

This type of scaling is useful as it assumes that people have a consistent view on a topic and they can be arranged in a hierarchy.

The Guttman scalogram was so widely used that researchers started to make experiments and used this system through surveys. After the birth of the Guttman technique, people waited 20 years to find out, visualizing it could be a hard task. Dennis Rondinelli basically grabbed this whole technique and visualized it in a heatmap to measure and summarize facilities statistics for settlements in 1980. That was one of the first heatmaps that not just used colors, but also arranged the data in a merit order.  

The exact moment when heatmaps entered its doors to the publicity

After the 50s, the tendency of improving shaded and colored illustrations had stopped. The main problem was the paper itself. The scientists believed that no matter how hard they try, most of the heatmap-related diagrams tend to be incomprehensible for a non-expert, so they usually just leave it out of their work. It is very ironic that nowadays, for example, a biological dissertation is usually full of heatmap-related images and diagrams. Some of us may think that the main cause of this increased popularity has to be the appearance of computers, which can analyze data and diagrams not just in 2D but in 3D as well. They are right, but only ½ of their opinion. Although heatmaps popularity has risen as the result of the appearance of computers, nowadays magazines and newspapers also use this format, as some of us realized how practical and easy-to-use it is. Moreover, a powerful heatmap is eye-catching as well, so it can increase the entertainment, but also the reader usually recognizes the data better than if it just contains written parts.

Did you know why we call it that way?

It is a very easy question to answer. In terms of heatmaps usually we indicate something commonly occurring with red color, blue means the opposite. However, there are other terms how researchers and writers used to call this analytical tool. Moreover, there are regions where people are not really familiar with this expression. Instead, they use the term “intensity map” or “point density interpolation”.  

The big break-through: computer-generated heatmaps

But what does a modern heatmap look like?  Luckily, the essence of the heatmap hasn’t changed through the years: it’s a method of representing data graphically where values are replaced by colors. Although nowadays computers edit them instead of humans, so it’s more precise and correct. 

Computer-generated heatmaps were developed in 1993 by Cormac Kinney and some of his colleagues. An interesting part of the story is that he trademarked the term “heatmap”, but then he unintentionally abandoned it. Just think about it: without this little mistake, we might not know and use this term nowadays.

Years after Kinney’s invention, businesses had not yet used his whole idea, although, in the field of meteorology and the stock market, it was a widely used implementation. Since computer-generated heatmaps could be used to visualize the results of numerical weather prediction models, allowing scientists to identify patterns and make more accurate forecasts.

10 years after Kinney developed computer-generated heatmaps, websites started to appear in the online world. There were, of course, business, e-commerce related sites, where there was a demand to somehow track users information, to understand their client’s behavior better. So there was only one little thing missed: the main supply part of this curve (if somewhere in the market there is a demand, supply will come!). Of course, these companies didn’t have to wait so long. There is shady information about that part of the story, but to the best of my knowledge, Crazy Egg was the first online site, which introduced and sold different kinds of heatmap analytics on their web page in 2006. 

Another big improvement in this area of life happened in 2013, when Michael Waskom introduced a Python package named Seaborn. Python is a general-used programming language, while the Seaborn package allows people to create different kinds of heatmaps if they are interested in this programming language.

The next trend and also evolution of what heatmaps have gone through are the release of interactive heatmaps started in the late 2010s. As of this, people who didn’t have a clue how it is working, now get access to it via these simplifications. A wide range of people have tried out this analytical tool in recent years, also numerous people have realized how easy-to-use and simple the heatmap itself is. 

Conclusion

Heatmaps have been with us for 150 years and as it looked at the beginning, we should say it has changed a lot. Its lookalike, the number of dimensions, the areas of usage; it all changed. But also throughout its evolution, there have been parts and elements which have remained stable. 

If heatmaps would have not improved through the years, there is very little possibility that we would use this web analytical method nowadays. Luckily, curiosity is a well-known human attribute, which we must thank a lot, as it always helps us to find the perfect solutions for different problems.

Don't forget, sharing is caring! :)