I created an activity in December/January for my students. I assigned it yesterday. In between, Desmos has altered the way that it presents the equation gotten from logistic regression (including what it names the constants).
We found this out when some numbers didn’t make sense to a student.
I think it might be an easy fix on the CL side (just some variable name changes), but once I do this, will those changes show up in the activity my students are doing (or would I need to cancel it and reassign it)?
Sorry this happened. It’s awesome when improvements are made to the platform but it is definitely frustrating when the changes mess up the functionality of a previously designed activity.
As to your question, I am not aware of a way to ‘push changes’ on a previously assigned activity. I believe you’ll have to archive/unassign the activity, make the changes you want, and then re-assign it.
Thanks. That’s what I figured about any changes I made, but wanted to ask just in case I had missed something.
What’s unfortunate about this is that the format of the equation they switched to is not what I’ve seen most mathematicians use. So it’s interesting why they chose it (one of those things when I’d love to be a fly on the wall to hear the discussion about why to go that direction).
Now to fix it, and hopefully not have to do a bunch of reprogramming.
Oh I think I know this one! The previous logistic form, c/(1+a*exp(-bx)), has a chance of being undefined for a bad guess of a and b (if b=0 and a=-1…). The new one (a/(1+exp(-(bx+c)))) avoids that problem, since exp is always positive.
As a general comment, it’s always going to be a little dangerous if you expect the students’ regressions to have the same variable names as you had when developing, since the calculator tries to be clever and avoid reusing variables. If the student types a=5 for some reason before exporting their regression, for example, it’ll name the regression parameter a_0 or something. I ran into that once upon a time, assuming e1 was going to exist in the graph and hold the list of residuals. Turns out I changed the variables from x1, y1 to x_{taco}, y_{taco} (or something) and the residual was actually being stored in e_{taco}
