Data is Wondrous
I don't know where to begin with this, but I'll try: I am weirdly passionate about graphs. The idea that you can order numbers in space and time, and make them make lines and curves that tell you things, is simple - you're probably used to making graphs (and possibly heartily sick of them). It's so simple that I think we forget how powerful graphs are, that they take millions of experiences and put them in a form we can understand more intuitively.
I'm going to put forward a proposal that will probably make most of you groan in frustration: graphs should inspire people.
Okay, when I write it that way, that just sounds confusing. Inspirational quotes are one thing, but what the hell is an inspirational graph?!
I can't really answer my own question, unfortunately, except to say that an inspirational graph moves you on an emotional level. It reminds you of something beyond the graph, or makes you feel like you had an epiphany, or gives you some quiet satisfaction.
That doesn't help either, does it? I'd bet that the better part of readers have never been emotionally moved by some bullshit bar chart or someone else's poorly put together pie chart. That's okay. 90% of graphs are crap.
Someone like me, who's more comfortable with numbers than people, actively seeks out and tries to create the 10%. For me, graphs are another kind of language (language doesn't have to be verbal).
You might be looking at that graph and wondering where the inspiration is. In fact, you might be looking at that graph and cringing at the colour scheme (the console generated it automatically - I didn't bother to set it). But this is a graph I find inspirational, and I'm going to try to explain why.
This graph is modelling the light curve for a microlensing event in the Andromeda Galaxy. Before I even started, I decided I only wanted to model the part of the light curve with the characteristic pixel-lensing "spike" - I was looking at data from the Liverpool Telescope during 2004-2009 and at that time the telescope wouldn't operate during summer months. This left me with huge patches in my light curve that made fitting difficult.
If you look at the red curve, that's supposed to model a pixel-lensing light curve. Notice how it is flat. Also notice how my data points (the green dots) curve upwards. That's from a periodic change in base flux which would have made fitting a pain in the posterior.
Here's where the wonder comes in: that blue curve models the baseline. After some unsuccessful tinkering, I found that I could quite happily subtract that baseline and get something lovely and flat. I could test my own data. There's the wonder: I could look at a galaxy millions of light years from our own and check microlensing events for myself.
This is how data is wondrous. Data should make everything fall together - in other people's heads as well as yours.
So make your bar charts and your pie charts and your graphs of your errors beautiful. Make them inspire people. Make them want to see what the graph is telling them. Make them want to see beyond that.
I'm going to put forward a proposal that will probably make most of you groan in frustration: graphs should inspire people.
Okay, when I write it that way, that just sounds confusing. Inspirational quotes are one thing, but what the hell is an inspirational graph?!
I can't really answer my own question, unfortunately, except to say that an inspirational graph moves you on an emotional level. It reminds you of something beyond the graph, or makes you feel like you had an epiphany, or gives you some quiet satisfaction.
That doesn't help either, does it? I'd bet that the better part of readers have never been emotionally moved by some bullshit bar chart or someone else's poorly put together pie chart. That's okay. 90% of graphs are crap.
Someone like me, who's more comfortable with numbers than people, actively seeks out and tries to create the 10%. For me, graphs are another kind of language (language doesn't have to be verbal).
This graph is modelling the light curve for a microlensing event in the Andromeda Galaxy. Before I even started, I decided I only wanted to model the part of the light curve with the characteristic pixel-lensing "spike" - I was looking at data from the Liverpool Telescope during 2004-2009 and at that time the telescope wouldn't operate during summer months. This left me with huge patches in my light curve that made fitting difficult.
If you look at the red curve, that's supposed to model a pixel-lensing light curve. Notice how it is flat. Also notice how my data points (the green dots) curve upwards. That's from a periodic change in base flux which would have made fitting a pain in the posterior.
Here's where the wonder comes in: that blue curve models the baseline. After some unsuccessful tinkering, I found that I could quite happily subtract that baseline and get something lovely and flat. I could test my own data. There's the wonder: I could look at a galaxy millions of light years from our own and check microlensing events for myself.
This is how data is wondrous. Data should make everything fall together - in other people's heads as well as yours.
So make your bar charts and your pie charts and your graphs of your errors beautiful. Make them inspire people. Make them want to see what the graph is telling them. Make them want to see beyond that.
Comments
Post a Comment