Menu
What's New
By:
Filters

Are you smart enough to found a solution???

D

Dr.D.C.

Elite
Joined
Jun 22, 2008
Posts
4,640
Location
Florida
Society
Glory Hound Irregulars
I need a function to extimate the tt value of skills. In entropiatool i found this table here than i ploted the values in excel 2007(but also in the 2003) i created a trend line polynomian order 6 with relative formula, in the picture



But OR this formula is wrong OR i m not able to re-formulate it in a way that excel can understand..... in effect when i try out some x values the relative y ones are far away from the aspected values.
There is around some guy smart enough to help????
thx in advance
DC
 
put Cat into the equation and you're good to go
 
Dr.D.C. said:
yup, it s very bad. Just a polynomial fit well from degree 4.
Click to expand...

I've never tried to fit this particular distribution, but I'm 100% sure you are looking for an exponential function.
 
you can snap the value in excel and then create a trend line exponential.....it will not fit very well.
 
what's you problem exactly?
type in your formulas and calculations here, and tell us what's wrong, so we could check
 
y = 6E-21x6 - 3E-16x5 + 4E-12x4 - 2E-08x3 + 6E-05x2 - 0.0543x + 12.34
this is what excel give me
 
humanwoman said:
and what's wrong with this one?
Click to expand...

that when you add the x values (skill points) it don t give you back the correct results y (tt value). or i m very bad to use it. Did you try it?
 
Dr.D.C. said:
that when you add the x values (skill points) it don t give you back the correct results y (tt value). or i m very bad to use it. Did you try it?
Click to expand...

There are many problems with this Excel polynomial thing, the biggest problem here I think might be the precision:
if your x is around 10000 and you gonna raise it to the power of 6, it's gonna be 24 zeros number, so your coefficients must have at least 25 valid numbers to give you back the result you are looking for

Excel is just not that good, you can set it up to display more numbers, but it won't display more than 15 digits for polynomial coefficients anyway:
http://support.microsoft.com/kb/282135

You can Google some alternative formulas like the ones here http://www.eng-tips.com/viewthread.cfm?qid=154948
or drop it and use some math packages, they should be better

P.S. I tried that table in my Excel and got coefficients with the maximum precision:


they won't reconstruct the original curve :(
 
Last edited:
humanwoman said:
There are many problems with this Excel polynomial thing, the biggest problem here I think might be the precision:
if your x is around 10000 and you gonna raise it to the power of 6, it's gonna be 24 zeros number, so your coefficients must have at least 25 valid numbers to give you back the result you are looking for

Excel is just not that good, you can set it up to display more numbers, but it won't display more than 15 digits for polynomial coefficients anyway:
http://support.microsoft.com/kb/282135

You can Google some alternative formulas like the ones here http://www.eng-tips.com/viewthread.cfm?qid=154948
or drop it and use some math packages, they should be better

P.S. I tried that table in my Excel and got coefficients with the maximum precision:


they won't reconstruct the original curve :(
Click to expand...

actually it don t looks bad at all but i checked the formulas and like usual them are not correct.....
i m trying now to approximate the data to an hyperbole using x axes like asymptote on the focus.
 
Last edited:
Learn to use vlookup.

Problem solved.
 
You must log in or register to reply here.
Back
Top