YARG!
For some reason I've been calculating covariance as...
Cov(X,Y) = E(XY) - (E(X)+E(Y))
Instead of...
Cov(X,Y) = E(XY) - E(X)E(Y)
Man I'm stupid. I've been beating my head against this wall for days, and this is review...
Currently Listening: Nine Inch Nails - Every Day Is Exactly The Same
Comments
Sweet, a math post. Do you know where this is derived from? That's the easiest way to remember.
Posted by: Nicholas | June 26, 2005 8:20 PM
No, the proofs for all this crap are so mind-numbing that the best bet is just to memorise them.
The real problem was that I was remembering it right, but for some reason I was adding the two means (E(X) and E(Y)) together instead of multiplying them. It doesn't make any sense, as it goes very specifically against the formula I had memorised.
As I said, I was being stupid :-/
Posted by: Kurrs
|
June 26, 2005 9:32 PM
There's a difference between the actual proof of this gabbledy-yak and understanding the motivation; the only way I ever got through the sheer mind-numbing BRAINROT that is 2nd year stats was by means of the Cartoon Guide to Statistics and by staring at the actual material until the whys make sense.
It would help if somebody actually wrote a decent book on this stuff. Future writing projects.
Posted by: Nicholas | June 26, 2005 9:56 PM
I think I'm going to have to stick to my "I was being retarded" party line because, when you think about it, covariance really makes sense.
Because the variance of one variable is:
E(X^2) - E(X)^2
So, the variance of two variables being
E(XY) - E(X)E(Y)
Is really quite intuitive. Instead of squaring, you're just multiplying the two variables together, ending up with a similarly useful but different purposed number.
And yes, a decent book would rock. A friend of my mom's is sending me some texts my way, she ran into the same problem we did and decided to go looking around. We'll see how well she fared.
Posted by: Kurrs
|
June 27, 2005 10:31 AM