rivka: (her majesty)
[personal profile] rivka
In the years since high school, I've managed to completely forget how to do algebra.

No, scratch that. It's just been in the years since I took the GRE. I know I could do algebra when I took the GRE.

*sfx: tearing hair*

Re:

Date: 2003-05-20 07:00 pm (UTC)
redbird: closeup of me drinking tea, in a friend's kitchen (Default)
From: [personal profile] redbird
That's one equation in two unknowns; lacking a known value for y, it's a curve.

Re:

Date: 2003-05-20 07:10 pm (UTC)
From: [identity profile] rivka.livejournal.com
Right. I was looking for a general equation in which to plug different values of y, to get x.

The specific y value I was working with was 0.246, but it doesn't really matter to the construction of the general equation.

(The values of y I have are odds ratios, which I want to convert to probabilities.)

Re:

Date: 2003-05-20 10:41 pm (UTC)
From: [identity profile] lerryn.livejournal.com
My solution to that is to plug in the *insert obscenity here* y value of choice, then worry about solving for x. I seem to recall the real solution invoving taking the log of both sides, but that's just a pain.

Re:

Date: 2003-05-21 06:18 am (UTC)
From: [identity profile] rivka.livejournal.com
I had the y-value-of-choice, and I still couldn't solve the goddamned equation. And I did very well on the Math GRE! It's inexplicable.

Re:

Date: 2003-05-21 03:44 am (UTC)
ext_16733: (Default)
From: [identity profile] akicif.livejournal.com
I cheated - rearranging the equation to give things like x = y + xy didn't really help, so I tried slotting in values of 0, ±1 and ±100 for x to get the general shape of the curve. When the absolute value of x is large y tends towards -1, when x is 0 y is 0 and there's a discontinuity at x = 1, where y effectively goes from plus to minus infinity.

This, of course, didn't help solve the beastie at all, so I tried sticking it in a spreadsheet and graphing it, then zooming in to give:



It is a cheat, and there's no doubt far better software for doing it with than Excel, but when the tool I've got's a hammer....

Date: 2003-05-21 05:00 am (UTC)
From: [identity profile] wcg.livejournal.com
Hey, it doesn't have to be that complicated, really!

To solve y = x/(1-x)

First, invert: 1/y = (1-x)/x
Now rewrite the right side of the equation as 1/x - x/x
Which is 1/x - 1

So now we know that 1/y = 1/x - 1
Or, 1/y + 1 = 1/x

From there just invert again, and you end up with:

x = y/(1+y)

Date: 2003-05-21 06:14 am (UTC)
ext_16733: (Default)
From: [identity profile] akicif.livejournal.com
<FX="slaps head" /> Of course! Forgot I could invert .... it's not just algebra I'm losing, it's basic maths.

Date: 2003-05-21 06:19 am (UTC)
From: [identity profile] rivka.livejournal.com
I forgot that I could invert, too. Unless I never knew it in the first place. Yay for [livejournal.com profile] wcg.

Date: 2003-05-21 06:37 am (UTC)
From: [identity profile] rivka.livejournal.com
Misha solved it in about ten seconds without inverting anything, so now I feel even more abashed. I mean, even if I did forget that you can invert, I should have been able to find this solution:

y = x/(1-x)

x = y(1-x)

x = y-yx

yx+x = y <--(this is as far as I got)

x(y+1) = y

x = y/(y+1)

Profile

rivka: (Default)
rivka

April 2017

S M T W T F S
      1
2345678
9101112131415
16171819202122
23242526272829
30      

Most Popular Tags

Page Summary

Style Credit

Expand Cut Tags

No cut tags
Page generated Jan. 19th, 2026 02:53 pm
Powered by Dreamwidth Studios