What’s this graph?
The cosmic microwave background? Preliminary results from the LHC? No, it’s SteelyKid’s feeding schedule. The horizontal axis is in days since we brought her home, the vertical axis is time of day on a 24-hour scale (in half-hour bins), and the color scale indicates the duration of the feeding in minutes.
If you can spot any clear pattern in this, you’re doing better than I am. There may be a hint of a developing pattern toward the end (she’s eaten at 1am, 8am, 1:30 pm, and 5pm each of the last four days), but there’s still a lot of scatter. There were some points in the first week when things looked fairly regular, too, but that fell apart pretty quickly.
We’ll keep tracking this, though, and it’ll be interesting to see if a real pattern emerges over the next few weeks…
Oh, like Trixie Update, you are charting upside down.
I totally see a pattern. There is a blank space on this graph somewhere after 22 hundred hours for more than half (the more recent half) of the time frame.
Obviously, someone is falling asleep at the job. A few more weeks, hopefully even more sleep!
Have you tried using 23 or 25 hours on the vertical, and binning those into “days”?
Coturnix: Oh, like Trixie Update, you are charting upside down.
In scientific graphs, measured quantities increase as you move up or to the right. What you stamp collectors do is your problem.
Tom: Have you tried using 23 or 25 hours on the vertical, and binning those into “days”?
To what end?
The way I’ve put this together (in SigmaPlot) makes it really difficult to do that sort of thing, so I’d need a really good reason…
I did think for a while that she was running on a 23-hour “day”– see, for example, the downward-slanting diagonal groups in the lower left– but I don’t think that really holds up.
We are all totally products of our subsection of science.
I cannot see any patterns at all, because my brain is only looking for vertical patterns. The horizontal? I can’t see them at all. Even after knowing which axis is which, my brain is refusing to see meaning.
I’ve been looking at dose-response and biomarker and other vertically arranged data for too long!
Can you post the swapped axes version in case somehow that does trigger something in us stamp collectors please? 😉
fft?
Plus I can spot several “lines” such as from (5,8) to (10,2), consisting mostly of 15-20 minutes of feeding, if I read the colorbar correctly.
In scientific graphs designed to show daily patterns of biological (including human) events, measured quantities increase as you move DOWN and to the right. It is called an actograph and that has been the agreed rule for a century. This makes it easier to see (for us Westerners at least) because this is the way we have been taught to read.
It does appear that feeding times are longer during the daytime than in the early morning, but the pattern weakens towards the last week.
Don’t tell me you haven’t learned the First Rule of Baby Behavior: Once you have figured out my patterns, I must change them. With you graphing the data, SteelyKid just has to be that much more random about it!
You won’t necessarily find a pattern until 5-6 weeks. Babies are born a few weeks before the brain is really ready to go – they’d be too big otherwise. There will come a day when you look at SteelyKid and will see that she’s looking at you with purpose, that a switch has flipped and there’s some real organization there. (At that point, just make sure she’s never awake longer than two hours and you’ll be stylin’.)
I spent a few days with my nephew when he was 5 weeks old. The first few days he was very like a newborn – disorganized movement, mostly-random staring, etc. The morning I left, I looked at him and could see he was different – actively looking around, definintely engaging different people, and so on. A few days later I checked up on his Trixie Tracker page and the difference was obvious. Starting on that day he had consistent sleep/wake/eat cycles. Totally “Oh, hello, your brain is now ready for use.”
Looks like a Hovmoller diagram (time-longitude) used to find east- or west-propagating atmospheric waves, which show up as diagonal lines.
1) Fourier transform to get time into the frequency domain.
2) Hold hardcopy at an acute angle, sight down the paper, and rotate. Linear correlations will appear as lines.
3) Print small B&W transparency, shoot an expanded collimated laser beam through it. Got correlation?
If Steelykid is a Boltzmann brain… publish! Then duck.
I was indeed thinking that there were indications of 23- or 25-hour “days” There are a number of diagonals, or at least hints of them.
Cool idea, but that’s a horrible way to view the data. Try a histogram or look at a Fourier transform.