So another situation I come across at work is nested designs. Kinda like repeated measures but doesn’t always have to deal with time. It could deal with space too. Kinda like babushka dolls, where a small doll is inside a big doll. And a smaller doll is inside that doll. And a smaller doll is inside that doll. And … yeah, you know how it goes. Or like when Randy tries to find Anton again, he realizes that cities can be nested within dimensions. And shops can be nested in those cities. And books are nested within some of those shops. And my tril … yeah, you know how that goes too. But anyway, how does Randy know in which shop in which city in which dimension Anton could be in? Well, it might depend on the variability within each of those levels … something we could cover next time. Until then …
Awe … little birdies in a nest! Cute, eh?? I’m improving! I hope.
So sometimes data don’t look like they look they follow one but follow two distributions. And look like this
Like good dimensions can follow one distribution and bad dimensions follow another. And how can Randy, for example, determine which of these dimensions he’s in based on the distribution. Well, for example, he could, um, well. he could, uh determine that based on the average number of outfits that Tina has in that realm. Okay, maybe that’s not the most exciting example, but there are many other exciting things that happen in both bad and good dimensions, promise! Just see for yourself! Like in one dimension, Randy chases some bad guys who may or may not know bad Anton through Las Vegas!
Yeah! Luck be a lady tonight! Okay! Yes, I will try again to do better next time. Until then — what stays in … you know 😉
So I just written up a prostate cancer paper on how correlations can be different in difference race subsets and why that’s important. And why is that important? Because you can have different interactions depending on the subgroup you’re in. For example, the number of designer dresses that Tina has could correlate with how many paintings her cousin, Jane, paints in unjustified dimensions but not in justified dimensions. How so, you ask? Well … Intrigued yet? What? Oh c’mon! Okay, I’ll try again another week. Until then, just imagine you’re Tina in an unjustified dimension and this is your closet…
Nope? Nothing? Okay, okay, I’ll try something different yet again then … just for you.
So I’ve actually attended the APS March Meeting in San Antonio this week and attended some cool talks on physics in medical research, physics and quantum computing, physics outreach, and you want to know one cool way that can help with outreach? Oh yes, you do! C’mon! All right. I’ll move on then. I also learned about something called the flat histogram. Which I will attempt to explain here. So here we go. Okay, you’re ready? Okay, me too. Lets start now! So it’s an algorithm where you keep resampling and update the variance, or how different the data points are from each other, until you look like you have a flat or uniform distribution. So lets say at the meetings, I could stay at Hotel A, B, C, or D, most likely staying in Hotel A, which I get when I sample the first time. But to add variation to my sampling, I decide to pick from only B, C, or D next time. And if B is selected next time, I select from C or D the time after that and so on. So our histogram will eventually become more flat like this. So then eventually I would have an equal chance of staying at Hotel C, say, as I do of staying at Hotel A. Hotel C, that is, where I just attended a staged physics play. Like a staged play of Order of the Dimensions would also be cool. With these potential candidates for my fantasy cast. With … yes, I’m stopping again. So I’ll just leave you with a pic of the River Walk I took earlier during my trip. The River Walk right by Hotel C where … yeah, okay, stopping now.