We filled up the last sheet on the legal pad we’ve been using as a baby-feeding log, which reminded me that it’s been a while since I updated this:
Again, this is the feeding pattern for SteelyKid, with darker colors indicating longer duration. Bottle feedings are arbitrarily assigned to 20 minutes for 4 oz of milk or formula.
If you can find a pattern in this, you’re doing a whole lot better than I am. In the last couple of weeks, especially, SteelyKid has stubbornly refused to establish any sort of pattern. She goes three hours between feedings during the day (more or less), but there are occasional four or five hour gaps that throw the whole thing off.
This makes it damnably difficult to schedule anything.
You’ve definitely got a pattern in there, look at the echoes of the peak around 75 days.
You could try some geophysics tricks. Convolving the data with itself and look for peaks at low times. If there’s just randomness, there won’t really be any. If there’s signal, it gets a lot more obvious — you take the data and fun a Wiener filter over it (assuming you have access to software that lets you do that without a bunch of work). Another would be bandpass filtering to remove frequencies below 1/(20 min) and above 1/(9 hour).
Basically, try the frequency domain.
If you squint and hold it sideways, you can see the Starship Enterprise.
I thought it was one of those Magic Eye dealies…
My first thought was that it was a still taken from the game of Life.
And, if you think about it, that makes a lot of sense.
I see a 4 hour cycle, with some noise in it.
The problem is that the drift forces delays into the future of the cycle, so it shifts day by day.
I recommend picking a 4 hour feeding schedule, and sticking with it. If she needs to eat outside the cycle, feed her a little bit, but you’ll probably be able to stabilize the cycle.
I don’t think we saw any signs of establishing a routine with our baby until the 6-9 month period. And then it was slow and would fall apart occasionally. Not that we were measuring as such but that’s my feeling. It didn’t bother us in any way.
It’ s much harder now she’s seven and our routines get kicked apart 2-3 times a year by jetlag : )
Have you tried a Fourier transform? The magnitude of the complex result should tell you something about her regularity, and when she shifts her pattern during one of those longer intervals, it will strongly affect the complex argument. You’ll have to unwrap phase, but that’s easy on 1-D data.
You DO have matlab, I assume?
what was the weather like? any overlays? more calories are burnt when it’s cold no? was it unseasonably warm on the days where there were the long gaps?
Could be scalloping, or split rhythms (one period longer than 24h, the other shorter than 24h). It starts entraining but then it breaks off into a freerun for a few days – a version of relative coordination?
I vote for split rhythms too….I was going to suggest 25 hours because that’s the natural cicardian rhythm, but SteelyKid seems like a complex individual….what would a 25/23 Y axis look like?
Yup, I’d say those are two rhythms, each showing relative coordination. Occasionally they fuse into one, at which point they can entrain – then they split again and each freeruns with its own period. Pretty normal for this age:
It’s a chevron pattern that shifts about an hour each day, masked by some noise. It started around Day 60.
It doesn’t look plain “random”: look at the clusterings of sequential blobs at 45 and 135 degree angles. Maybe that’s just an artifact of sampling/representation procedure and isn’t relevant, but I note the observation since such issues weren’t given AFAICT.
BTW Chad, good luck with the baby and I am impressed that you can also meet your professional obligations as well as post and answer comments (and I am again autochastened over my recent peskiness – will be happy to see your thoughts on the QM collapse problem whenever you can or do get around to it. I’ll also try to check into your book.)
PS: Heh, recounts are in the offing in Minn. but I don’t think physical issues about ballots will be the factor, sorry but I just can’t avoid that certain subject when I see your name. It’s like “Madonna” started in the 80 to mean the singer instead of the mother of Jesus 😉
Ummm, I mean I can’t avoid thinking about that subject of vote counting when I see “chad”> So I’ll avoid talking about it …
Are you using Mathematica in analyze the baby? A New Kind of Baby Book!
These sequences of four or five dots along diagonals (of the same slope!) are unmistakable. What’s the explanation?
“This makes it damnably difficult to schedule anything.”
That’s the point. You are not supposed to have a life.
Just wait till she really gets the hang of it; this is just practice.
🙂
I think this is just some oblique, lower-dimensional slice through a higher-dimensional surface that DOES show a pattern.
What it is, I don’t know. If you had taken her temperature every 2 hours or so during the data collection, you might have seen some sort of Temperature/Metabolism/Hunger correlation. Who knows. Maybe there are 10^500 different parameters, and each timestep is really half of the Planck Time.
Other than that, I’m pretty sure I see Jesus riding a Jetski in the lower left-hand corner.
Who’d want to schedule anything when you’ve got this amazing baby producing all this amazing data……
Wait until she walks.
You kid is often functioning in 5 hour feeding intervals independent of the time of the day/night, so you see approximately 1 hour drift in the feeding times every day. After few days of a nicely linear drift – just to make things more complicated – the kid suddenly switches to 4.6 hour cycle; which then causes the feeding time to drift back for a while by about hour a day. Chevron pattern, as the previos commenters noted.
This is so awesome. Finally, the reason the internet was born. I always think about data like this, but never have the stamina to collect it all. Good job.
Try changing the y axis to 24 hours, 39.5 minutes. I think you’ve got yourself a Martian baby there.
I just fed my kid during class (including physics class and finals). It worked great.
That drift has nothing to do with Mars. Somewhere, sometime long ago, I read about a study where they put people in a cave with no time cues. They lived on a 25 hour clock. Anyone who travels knows that it is easy to stay up one hour later every day, just as “fall back” is easier than “spring ahead”.
The drift is quite plausible, and the reset is probably parental or diurnal cues.
Predictable feedings? For babies? Pshaw! I know there are those who claim to have a baby who is, but I suspect that about half of them are lying. The other half are worthy only of our envious scorn – at least our babies are cuter than theirs.