line of best fit/reading values from graphs

[Guest]

i have excel 2003 and am trying to create graphs for an electronics lab write
up. I have entered information into a table and created scatter diagrams etc
but i want to put on a line of best fit. The trendline function does not work
well enough for what i want, and i remember there being an option for a line
of best fit on an older version of the program. is there a way to do it? The
reason i am asking is because i want to read off values from my graph, i.e.
on a voltage vs frequency graph i want to know what the frequency would
theoretically be at 0.707 volts. if i can not create a best fit line, is
there a way to do this??

 

[Bernard Liengme]

You did not tell if the fit is linear or not
You can use LINEST to fit to linear, polynomial, log etc
See www.stfx.ca/people/bliengme/ExcelTips and
www.tushar-mehta.com

 

[Guest]

they're not linear or polynomial. some are like exponential but upside down
(yet not logarithmic, dunno what they're called). some are like the shape of
a black body curve if you ever did physics alevel. they're weird shapes that
can't be mapped onto a 'normal' function. what i really need to know is how
to get excel to read x values for a given y value

 

[MrShorty]

A couple of ideas:

1) Are your data spaced closely enough to use interpolation instead o
curve fitting to get reasonable results? Excel doesn't have a built i
interpolation function (some other spreadsheets have built in linea
interpolation functions if you're not tied to Excell), but it can b
done in Excel. Either write a UDF for it, or work out the looku
portion using INDEX(MATCH()) then interpolate on the located interval.

2) If you insist on using a curve fitting approach, you've got to com
up with some kind of function that will represent your data. Trendline
and LINEST really only work on linear parameter (or equation where th
parameters can be linearized), but there are other functions that don'
use linear parameters for which the parameters can be found using
numerical approach like Solver. For example, you mention that some o
your data look like a blackbody radiation curve. Been a long tim
since I dealt with that stuff, but a quick Google search found that
blackbody curve is an equation of the form y=(Ax^-5)/(B^1/x-1) where
and B are constants (B is a function of temperature). In your case,
would just treat A and B as parameters. Can you come up with a fe
equations that will fit your data

 

[Bernard Liengme]

A curve fit with Solver would be possible if you have an idea what function
you wish to fit the data to.
Contact me privately if you wish to explore this idea.