I'm having some trouble with Excel's trendlines. I plotted a set of data and had excel plot a trendline to it. I then use the equation it spit out and it gives me completely wrong numbers. I plot these wrong numbers and the trendline for them and its the exact same as before.

What the fuck? Am I missing something completely obvious?

1/19/2009 2:36:18 PM

yes

1/19/2009 2:36:50 PM

you need to divide by 0 at the end

[Edited on January 19, 2009 at 2:43 PM. Reason : (yada yada)/0]

1/19/2009 2:43:09 PM

your 2E-07x^2 is rounded, leaving some digits out.

that's the only explanation i can think of.

1/19/2009 2:50:17 PM

When I have temperature in C rather than K, it gives an equation that works.

This is retarded.

1/19/2009 2:53:39 PM

What are the data points?

1/19/2009 2:56:14 PM

T ©	T (K)	     v (Pa*s)
0	273.2      0.001786
15.6	288.8	     0.001131
26.7	299.9	     0.00086
37.8	311	     0.000682
65.6	338.8	     0.000432
93.3	366.5	     0.0003066

1/19/2009 3:05:06 PM

urge is right. If you set the font number of the trendline equation to teh number category, with like it cranked ot like the 10th decimal place, you get y = 0.0000002314x2 - 0.0001626550x + 0.0288851725. I checked it with the Kelvins and got teh right other doodad.



Probably just shows up here because you're graphing big numbers against really small numbers.

[Edited on January 19, 2009 at 3:21 PM. Reason : .]

1/19/2009 3:20:10 PM

yeah, you are going to have to apologize to Lord Excel... its the greatest program ever

1/19/2009 3:28:48 PM

Thanks. I didn't think that the rounding would throw it off that much.

1/19/2009 3:47:56 PM

well...if you blow the scale up to normal numbers, it doesn't throw it off that much

1/19/2009 3:54:34 PM

look at your R^2 value, not even a difference of .02

just like ^ said, it didnt throw it off that much. you're just playing on a very small scale.

1/19/2009 4:06:17 PM

True, but using a negative viscosity to calculate Reynold's number kinda makes your data look weird

1/19/2009 4:17:06 PM

i only know excel.

i sure don't know what the hell you were just talking about though.

1/19/2009 4:18:31 PM

just make the graph in pc paint

[Edited on January 19, 2009 at 4:19 PM. Reason : n]

1/19/2009 4:19:17 PM

What is that, CE 382?

1/19/2009 4:21:40 PM

did you get it working? a lot of times when i'm dealing with graphs like that, i right click and check the cells that its linking to. i find that a lot of the times the cells aren't linking all the way and need to be moved to show all the data.

1/19/2009 4:46:23 PM

be sure to use x-y scatter plot and not line plot. using line plot will screw up your trend line with the quickness. i don't even know why line plot is an option.

1/19/2009 5:50:19 PM

Its for CHE 330.

Yeah I never use line plot, cant really think of when I would need it.

1/19/2009 6:04:31 PM

Quote :

"i don't even know why line plot is an option"

sometimes you wanna show how something changes over time

1/19/2009 6:13:59 PM

if that's the case i'll have a time array in my data set.

1/19/2009 6:16:47 PM

oen thing's for sure: excel makes some ugly-ass plots (unless you work pretty hard to de-ugly them).

1/19/2009 6:17:02 PM

not as ugly as matlab plots

1/19/2009 6:19:46 PM

i guess
my tecplot plots look far better with much less tweaking.

1/19/2009 6:20:48 PM