So I know I promised to talk about regression parameters but you guys! You guys!! I got an interview with Dr. Paul Halpern, a real life physicist on the multiverse stuff!! And here it is on my google page! So hope you check it out and check back next week 😉 See that? Physicists and statisticians can work together! My trilogy has hope yet! And Warner Brothers … okay, yes, I’m stopping again.
So how do we know the probability that we’re in a good dimension given a bunch of other factors? Well, we can for example check Tina’s marital status, employment status, the climate of the place where she lives, etc. and multiple them by these things called regression coefficients (something we’ll cover yet another time) and add them up together to get a number say, y. Then we take the exponential of y, to get, say, exp(y). Something cool to wiki. And then our probability is exp(y)/[1+exp(y)]. Voila!! So again, how the heck do we get y?? Ah … like I said, we’ll cover that next time! Until then, have a good week — oh, and if you still need something for dad this Father’s Day.
So as I mentioned last time, we might have something called a logistic regression model using logits as the response and how can I explain that … hmmm. Well, I can take the probability of being in a bad dimension, p, and then divided that by a number that’s one minus that probability (or 1 – p) and take the natural logarithm as explained here and voila, that’s our response. See, that wasn’t so bad! Right? And what’s the purpose of this? Well, it’s to map a binary variable, or a variable only having two levels, like whether you’re in a good or bad dimension, to a continuous variable. And how do we know the probability, p, say, that we are in a good dimension? Well, that depends on the factors we adjust for in the regression of course! And now, I have a subject to cover next time!
Until then — again, if you need some more summer reading …
So sometimes we model stuff to see how said stuff correlates with other stuff. But what if even more other stuff confounds that correlation. Well, that’s why we can put in that other other stuff in our model too. Which we talked about before. So we can put that stuff into the model. Like, say, Randy does find Anton in a warmer climate. Other factors he can adjust for could be Anton’s occupation, marital status, where he vacations and so forth. So if we put all this stuff in the model — such stuff are usually called covariates, by the way — we can get a better sense of which dimension they’re really in, making more sense than just going by the climate. But what kind of model are we putting all this stuff into anyway? Well, 1. — it’s called a logistic model and 2. we’ll cover it next time. Until then …
Just wondering … remember, summer just started!