[Regretfully, the web application discussed here has been lost in a botched site migration. Apologies. -Editor]
My goal is to introduce our Time Series Explorer tool.
I wrote the tool as a small demonstration of the power of interactive visualizations. I wanted to find data that was easy to relate to, and show how a single interactive visualization could allow somebody totally new to the data to quickly absorb the meaning of the data into their minds.
The whole point of data visualization is to efficiently put data into minds. Consider the input channels available to most minds: touch, smell, sight, taste, and hearing. Which of these is the highest bandwidth? Sight, right? It’s no coincidence that we encode data visually: it makes use of the highest bandwidth channel available to most of us: our sight.
That’s the input side, from the data into the mind. What about the output side? How can we get things out of the mind? And, for that matter, what do we want to get out of a mind?
Answer: decisions. Desire. The will. When the mind receives the data, if there’s a desire to understand, that desire can often be broken down into questions it asks. For example: What’s that weird thing here? Why is that line doing that? What happened last summer?
How do we express those questions to the data set? This is where interactivity comes into play. If you include the “right” set of controls into the visualization to let it redraw itself in certain ways, answers to many of the questions one might ask become apparent by zooming in here, focusing that line, etc.
So: visualizations are for inputting data into the mind. Interactivity is for outputting the will from the mind to the visualization.
Put the two together, and you have an integrated system of mind and machine, that can enable a mind to absorb the meaning of the data much more quickly and completely than, say, a set of static visualizations. (The idea, by the way, of an integrated mind/machine system is a major emphasis we have at Altometrics, and the big idea behind intelligence amplification techniques a la Douglas Engelbart and Bret Victor.)
(Note that having a so-called integrated system isn’t enough. It has to be a good one. Briefly, this means that the controls it exposes are intuitive to use, that the the system as a whole is quick to respond to changes in the controls, and that the set of controls available are able to answer the sorts of questions that a user might ask. Although I can imagine better systems for this data, I think that our demo here does a good job.)
Here’s a little about the data. This is daily high temperatures for a period of 13 months for about a dozen cities, including some international ones. The temperatures are in Celsius.
Anyway, try out the tool yourself and see if you can figure out what each control does. See how easily your mind absorbs the data.
There are a few key observations that I’ve noticed from the data. See if you can find them all. If you’d like to check your answers, you can jump to the bottom to see them
This dataset represents just shy of 5,000 data points. And it took you ~5 minutes to get a pretty solid understanding of what the meaning is here, right? You didn’t have to understand any complex models. You didn’t have to wade through a stack of visualizations. You could just dive in and start exploring—at least, I hope that was your experience.
Look for more things like this from Altometrics in the future. This is just the tip of the iceberg in intelligence amplification through data visualization. Data visualizations you can explore. Data visualizations with presence.
Here’s to an exciting journey!
- Berlin has what looks like a measurement outage for April and May of 2014. The straight line is the tell-tale signature of missing data points.
- There's broad agreement that it's colder in the winter months and warmer in the summer months.
- The coldest cities are Detroit, Denver, Berlin, Boston, and generally the more northern ones.
- The warmest cities are Miami, Sydney, Austin, San Diego, and generally the more southern ones.
- Two cities by and large buck the seasonal variation trend: Miami and Sydney.
- Miami is fairly flat on the whole, which, as Miami residents can tell you, is why it's a nice escape for people up north during the winter.
- Sydney seems to be anti-correlated with the other cities. When they get colder, it gets warmer, and vice-versa. When you think about it, this makes perfect sense: Sydney is the only city in this dataset that's in the southern hemisphere.