What is Visualization?
The American Heritage Dictionary defines visualization as:
(1) The formation of mental visual images.
In the context of visualization with data, it is necessary to add something to this definition, so that it becomes:
The formation of mental visual images to convey information through graphical representations of data.
If you are pursuing a career in data science, this is one of the most crucial skills that you can master, and it is transferable to virtually any discipline. Let us imagine that you are trying to convince your manager to invest in a company and you present them a spreadsheet full of numbers to explain to them why this is such a good investment opportunity. How would you respond if you were the manager?
If presented in visual form, information is often much easier to digest, especially if it makes use of patterns and structures that humans can interpret intuitively. If you want a quick and easy visualization that requires little to no effort, you can go with something like a pie chart or a bar chart. In terms of developing visualizations, this is often as far as most people go, and often, it is as far as they made need to go for there field of expertise.
Another factor that inhibits our use of visualizations is the amount of data we have available. How do I know if visualization is an appropriate method to communicate a message?
This is a difficult question to answer. One design study recommends that we assess the viability of using visualizations based on the clarity of our task and the location of the information.
If we are in the top right corner of this diagram, it becomes feasible to develop and program interactive visualizations, which is the realm in which data scientists are now entering due to the persistently increasing scale of data resulting from the information explosion.
We are now living in a data-driven world, and it is only likely to become more data-driven. This is clear from multiple areas, such as important advances in developing large-scale sensor networks as well as artificial intelligence agents that interact with the world (such as self-driving cars).
In a world where data is sovereign, having the power to develop clear and impactful visualizations is becoming an increasingly necessary skill.
Good and Bad Visualizations
Humans have been creating visualizations for thousands of years, and whilst the drawings of cavemen are slightly less spectacular than what we have nowadays, it is still good to appreciate just how powerful some of the early visualizations were, as well as how impactful they have been on the modern world.
Take Leonardo da Vinci, for example, an Italian polymath who was not only the first one to come up with incredible inventions such as the airplane, helicopter, and tank, but was also incredibly skilled at drawing. His engineering and anatomical drawings, like the ones below, are incredibly realistic and yet simple to understand.
Being skilled at drawing was very necessary for the purpose of visualizations back hundreds of years ago when we did not have computers to draw things for us. Take a moment to admire Galileo’s sketches of the moon during different phases of the lunar calendar.
It is not often that we really stare at ancient drawings of the Moon, so is there really still a need for these types of visualizations in the modern world? And if there is, can we not just leave it to artists, graphic designers, and the like?
The answer is obviously yes. Even ten or fifteen years ago, learning something like chemistry was incredibly difficult, despite being able to picture molecules in your head, it is still tough to translate between complex scientific words and your mental picture of what is occurring. Nowadays, one can go on Youtube and type in a few words and watch a visualization or a visual walkthrough of essentially any aspect of chemistry. This same idea applies for essentially any abstract idea in science.
So now we have convinced ourselves that visualizations are pretty useful for conveying information, and can also be used to explain complex ideas in a more interpretable manner.
What are some examples of good visualizations?
I currently live in Boston, so several of the following visualizations are related to the city of Boston. These are just some visualizations that I consider good, and, due to their subjective nature, you may disagree with me.
In Boston, we have an underground subway system called the T. As with any city subway system, there are a bunch of different lines and they go in various directions, and some of the lines take longer than others due to longer distances.
The below visualization captures not only the time taken to each stop from the city center in the form of concentric spheres but also follows the correct direction for each line. Looking at this diagram, it is pretty quick to work out which line to take, which direction it goes, and how long it will take to get there.
(1) The formation of mental visual images.
In the context of visualization with data, it is necessary to add something to this definition, so that it becomes:
If you are pursuing a career in data science, this is one of the most crucial skills that you can master, and it is transferable to virtually any discipline. Let us imagine that you are trying to convince your manager to invest in a company and you present them a spreadsheet full of numbers to explain to them why this is such a good investment opportunity. How would you respond if you were the manager?
If presented in visual form, information is often much easier to digest, especially if it makes use of patterns and structures that humans can interpret intuitively. If you want a quick and easy visualization that requires little to no effort, you can go with something like a pie chart or a bar chart. In terms of developing visualizations, this is often as far as most people go, and often, it is as far as they made need to go for there field of expertise.
Another factor that inhibits our use of visualizations is the amount of data we have available. How do I know if visualization is an appropriate method to communicate a message?
This is a difficult question to answer. One design study recommends that we assess the viability of using visualizations based on the clarity of our task and the location of the information.
If we are in the top right corner of this diagram, it becomes feasible to develop and program interactive visualizations, which is the realm in which data scientists are now entering due to the persistently increasing scale of data resulting from the information explosion.
We are now living in a data-driven world, and it is only likely to become more data-driven. This is clear from multiple areas, such as important advances in developing large-scale sensor networks as well as artificial intelligence agents that interact with the world (such as self-driving cars).
In a world where data is sovereign, having the power to develop clear and impactful visualizations is becoming an increasingly necessary skill.
Good and Bad Visualizations
Humans have been creating visualizations for thousands of years, and whilst the drawings of cavemen are slightly less spectacular than what we have nowadays, it is still good to appreciate just how powerful some of the early visualizations were, as well as how impactful they have been on the modern world.
Take Leonardo da Vinci, for example, an Italian polymath who was not only the first one to come up with incredible inventions such as the airplane, helicopter, and tank, but was also incredibly skilled at drawing. His engineering and anatomical drawings, like the ones below, are incredibly realistic and yet simple to understand.
Being skilled at drawing was very necessary for the purpose of visualizations back hundreds of years ago when we did not have computers to draw things for us. Take a moment to admire Galileo’s sketches of the moon during different phases of the lunar calendar.
It is not often that we really stare at ancient drawings of the Moon, so is there really still a need for these types of visualizations in the modern world? And if there is, can we not just leave it to artists, graphic designers, and the like?
The answer is obviously yes. Even ten or fifteen years ago, learning something like chemistry was incredibly difficult, despite being able to picture molecules in your head, it is still tough to translate between complex scientific words and your mental picture of what is occurring. Nowadays, one can go on Youtube and type in a few words and watch a visualization or a visual walkthrough of essentially any aspect of chemistry. This same idea applies for essentially any abstract idea in science.
So now we have convinced ourselves that visualizations are pretty useful for conveying information, and can also be used to explain complex ideas in a more interpretable manner.
What are some examples of good visualizations?
I currently live in Boston, so several of the following visualizations are related to the city of Boston. These are just some visualizations that I consider good, and, due to their subjective nature, you may disagree with me.
In Boston, we have an underground subway system called the T. As with any city subway system, there are a bunch of different lines and they go in various directions, and some of the lines take longer than others due to longer distances.
The below visualization captures not only the time taken to each stop from the city center in the form of concentric spheres but also follows the correct direction for each line. Looking at this diagram, it is pretty quick to work out which line to take, which direction it goes, and how long it will take to get there.
