graphspergers - the graphs and quantitative visualization thread

lol IOP... they're taking our jobs! I work for the UofC Press, on peer review and production. My old boss just accepted a position at IOP in the states, I'm hoping to get headhunted by him eventually :D

You're probably right about the figures, but I'd rather not chance it. It'd be kind of a shitty thing to bring down the credibility of the AAS!

I was on peer review and development ("publishing editor") for Journal of Physics A. IoP in the US is in Philadelphia, right? I never got a trip there out of that job, but I did get ones to San Francisco, Pisa and Bangalore, so it wasn't all bad.

As far as I know they're mostly in Philly, but the story is a new office is opening in DC, which is where my former boss is going.

Think I can get away with this one, though I am no physicist so I have no idea what it's about beyond pretty colors...

http://farm2.static.flickr.com/1122/653521560_09b57a1062.jpg

Isn't that the Great Wall/Coma Cluster thingy on the Map Of Universe jpg?

Might just be the Orion Nebula, mind.

analysis of stereolab riffs:
http://img124.imageshack.us/img124/6635/labrc1.gif

It's probably a galaxy cluster. Contours are probably the same object at another wavelength, e.g. X-ray, which is a good thing to look at clusters in. Distance scale is much too small for the Great Wall and too large for a nebula.

The graph I have been futzing with for two weeks. I want all 30 lines to flat and at y = 0.0.

This is what it looked like on Monday:

http://img170.imageshack.us/img170/9071/residualszp0.gif

Tuesday:

(link removed)

Today:

(link removed)

I'm here to tell you that this is the scientific enterprise. It's some bullshit.

What is your model made of? What is it supposed to be used for when it works?

I make models of disk galaxies. They are ones that happen to be oriented edge-on to us so they look more like http://en.wikipedia.org/wiki/Sombrero_Galaxy (but even more edge-on) than http://en.wikipedia.org/wiki/Messier_81. This makes it harder to work out whether they have spiral structure or bars than ones that are face-on, which you can just look at.

The idea is I take an image of the edge-on galaxy, add up the mass along the line of sight and predict how the stars in a perfectly asymmetric galaxy with that mass distribution would move. I then compare this to measurements of the stars movement (I will show these soon). If I can't get, e.g. model and data velocity-radius curve to match in the central region then I can conclude that the galaxy is not perfectly axisymmetric and probably has a bar. Another product of this fitting process is the mass-to-light ratio. If M/L > 1 then we either have stars heavier than our sun for the amount of light they produce -- or dark matter.

That is the big idea, anyway. In practice I end up doing things like what I've been up to for the past fortnight, which is basically dealing with a shortcoming of my data, which is good in some regards, but poorly calibrated. This involves futzing around with numbers until my lines match the lines for the same objects published by the big, very accurately calibrated surveys.

(link removed)

^^^ example of comparing my the velocity curve of my mass models to the data. That's a pretty good fit. M/L is worryingly high, but I know why that is. I need to regenerate all these plots.

What do you use to make your figures, caek? I do a lot of work on graphics, mostly because it seems there is a sizable group of authors using plotting software that they wrote themselves, and the fonts and/or colorspace get messed up. Also, IDL can eat my balls.

I use IDL : (

Ahahah I thought so! :) The big thing is that in my experience, IDL does not know the difference between RGB and CMYK. It may say it does, but it doesn't.

Pretty much all astrophysicists do. There are movements toward Python, mainly coming from the Hubble dudes, but there is an awful lot of entrenched code. I'm not convinced Python's graphics capabilities will make journals' lives a lot easier anyway, although it is a much nicer language.

I'm going to redraw my figures for publication in Illustrator. It's what Tufte would do.

Yeah, it has a very, very eccentric colour system, although in 1993 it was pretty amazing.

It is presumably why my pure white backgrounds come out off-white when converted to a GIF by Preview.app.

Working with your stuff in Illustrator is a good idea. Convert those fonts to outlines!

http://www.rotovibe.com/images/pill.gif

http://img.photobucket.com/albums/v417/albaalba/ilx/Picture1-1.png

Last three months' listening from LastGraph

(link removed)

So we signed off on those flat lines, and now I have to do error analysis (to figure out the error in my correction). This white line on this graph (not mine) is my life right now:

(link removed)

(link removed)

The other thing I am doing at the moment. The thing in the pink circle is one of the galaxies I am interested in. The big star to its right is going to ruin my day tomorrow.

More rap graphs:

http://www.jamphat.com/rap/

For the record, the original Last.fm graphs were made by Lee Byron (http://megamu.com/lastfm/ and http://centripetalnotion.com/2007/05/18/19:24:07/)

http://i22.tinypic.com/29ks9ys.jpg

surely the average height of the WTC would be lower in 2001, if measured monthly over the whole year.

Well, that's obviously Fiscal Year 2001 (trade center, get it?), that ended on Sept 30th. That is why the difference on such a small graph is almost unnoticeable.

http://img20.imageshack.us/img20/3050/residualsjr9.gif

lol science

http://img155.imageshack.us/img155/821/picture1cp6.png

That, my friends, is discretization and it's a bitch.

that hurts to look at

it hurts my soul.

http://i50.photobucket.com/albums/f325/caek/Picture1-2.png

That is a weird way of looking at my model galaxy. Smooth.

And here are a couple of graphs I did for a film (supposed to be bad)

http://i50.photobucket.com/albums/f325/caek/graph.png

http://i50.photobucket.com/albums/f325/caek/pie.png

And I forgot to post this when I started this thread, which you'll have seen if you're a fan of uppers

http://i50.photobucket.com/albums/f325/caek/speed3.gif

http://blog.iso50.com/?p=1024

http://blog.iso50.com/wp-content/uploads/2008/02/burkina.jpg

http://blog.iso50.com/wp-content/uploads/2008/02/somalia.jpg

way to undermine your credibility, dude

I made a sweet bubble plot today!!!

jpgs or it's not true

does anyone have any more graphs that look like vajayays, pls email me thx

i have some! i made these.

RAINBOW:
http://i50.photobucket.com/albums/f325/caek/13.png

STD GAMMA II
http://i50.photobucket.com/albums/f325/caek/13-1.png

which scheme do you prefer? they both look fine printed b/w, which is one concern.

Some group placed little white flags on our quad, each for "at least" 5 Iraqis dead, and little red flags, each for "at least" 5 Americans dead, from the Iraq war. The red flags are mostly clustered in a tidy area in front of the sign with our school's name on it, and the white flags stretch on for several blocks. It is pressing my art buttons, my math buttons, and my social justice buttons all at the same time.

Here is the first flickr photo I could find of it.

This may be old news, but wau Songchart

Number of soldiers riding away, grouped by metallurgical composition
Effect of Time on Perceived Relative Distance of Troubles

What kind of chart would you use to show, say, technological lag?

eg. you want to show, visually, a timeline of when the 486 became standard, then the pentium, then the p2... and then show when your company adapted the 486, ten the pentium, etc... and dsplay how long each took?

and, how the fuck do i make excel make whatever you recommend?

a personal favourite:

http://base58.com/ilx/opinionsvsrelevance.gif

xp: opportunity to procrastinate, yay. Two options.

One is to plot the year of release against the year of adoption. The further above the line y = x the points are, the later it was adopted. You'd probably label the points with, e.g. "Pentium" or whatever rather than just put crosses.

http://i50.photobucket.com/albums/f325/caek/Picture1-3.png

Another is to plot the lag before adoption (i.e. adoption - release). The further above y = 0 the later the adoption.

http://i50.photobucket.com/albums/f325/caek/Picture2.png

does this guy know how to party or what

you, sir, are made of heroism.

i think i am going to go with the first one, btw... just figuring out how to make it. any hot chart-makin' tips?

Uh, don't use Excel.

Sorry. Actually, that should be pretty easy in Excel. Set up three columns of data:

1988 1998 1988

1992 1993 1992

1994 1995 1994

1995 1998 1995

1998 2000 1998

2003 2007 2003

Then plot column 1 (release) vs column 2 (adoption) as points and column 1 vs. column 3 (release again) as a dotted line.

That may be a retarded way of doing it in Excel. I haven't used it since the Pentium II.

http://b3ta.com/challenge/graphs/

loooooooool

An analysis looks at how defense spending among the nations with the highest expenditures has changed since 1992 and what may have driven the changes https://t.co/3ln08vOKAo pic.twitter.com/yqK6MqwQUm

— St. Louis Fed (@stlouisfed) January 22, 2023

https://i.redd.it/cxtoiiuy9l9b1.png

see you there in April!