Ok, I just did up the cleric saving throws and have
some interesting results. First of all, the only
two types of equations that yielded consistently
high R^2s are linear and polynomial (I only used order
three to maintain sanity :)
The polynomial yielded better fits, but not usually
by a lot.
For example, breath (x-axis = level, y-axis = saving
throw):
y = -1.4788x + 80.621 R^2 = .9908
y = -.0012x^3 + .0664x^2 - 2.4745x + 84.082 R^2 = .9952
So, since the linear equations are much easier to use
and program, I will put forward the linear equations
and ignore the polynomials.
In fact, the coefficients for the equations for most of
the [cleric] saving throws are sufficiently close that
we can probably reduce all five to one single equation:
Para: y = -1.6033x + 52.218 R^2 = .9599
Rod: y = -1.4788x + 70.621 R^2 = .9908
Petri: y = -1.4785x + 65.717 R^2 = .9916
Breath: y = -1.4788x + 80.621 R^2 = .9908
Spell: y = -1.4785x + 75.717 R^2 = .9916
I would tend to reduce this to one equation:
save: y = -1.5x + 70
It is reasonably close to all the individual equations
(only on para is it really off), and it fairly trivial
to program.
The polynomial equations also tended towards one
single equation:
y = -0.0016x^3 + 0.09x^2 - 2.9x + 74
or, even more simply:
y = -0.002x^3 + 0.1x^2 - 3x + 74
I expect the other classes will all show this same
tendency. Do you all want one equation? Five
equations? Linear? Polynomial?
I'd like to know what direction I should go before
crunching the rest of the number ;)
---
eglamkowski@angelfire.com
http://www.angelfire.com/nj/eglamkowski/null.html <- Null webring
http://www.angelfire.com/nj/eglamkowski/eia.html <- EiA webring
Angelfire for your free web-based e-mail. http://www.angelfire.com
+------------------------------------------------------------+
| Ensure that you have read the CircleMUD Mailing List FAQ: |
| http://democracy.queensu.ca/~fletcher/Circle/list-faq.html |
+------------------------------------------------------------+
This archive was generated by hypermail 2b30 : 12/15/00 PST