So the formula the trendline formula seems to be returning R² with one method with an exponential formula and a different method with a polynomial method.
#Excel trendline equation precision series
Similarly I calculated a series of Ys with this formula, however this time comparing Ys I return the same value for R² = 0.952370227 I added a 2-order polynomial and the trendline formula returned I calculated a series of Ys with 22.9134805762*Exp(0.0168832568x), and comparing these Ys with the original Ys I also get your value R²=0.9394636 using =RSQ() With your data I replicated your R²=0.9302848013 in the formula shown on the chart. Unfortunately you can't include file attachments but you can of course upload a file to any file sharing site, eg OneDrive (make sure it's shared) įor your earlier question, you can include links and images using the icons above the frame where you post. A hypothesis you would have clearly dismissed before. With this information you could almost believe that your data indeed follows an exponential pattern. You will get a better approximation, and an even better R² (0.941603). Second, the above formula is not optimal! Try instead:10.994*Exp(0,0182311x). Insert a diagram and an exponential trendline with formula and R² for these data. If the exponential trendline however is systematically wrong or unreliable, that feature of Excel looses value.
If you want to know whether your data follows an exponential pattern a*Exp(b*x), then Excel should be a help. The diagramm displays an exponential equation all right, but an R² from a different equation (the one from the linear trendline to (x,ln) data)! If you take Wing's sheet and insert a column with a*Exp(b*x) values (a and b from the formula in the diagram) and then calculate the R² (with the worksheet function!) between this column and the Y column, you will arrive at R²=0.989. If you, however, cannot rely on what you see in an exponential trendline diagram. R² in this context is an important indicator: if under 0,95 you should forget the corresponding trendline (for the moment at least). Probably follow an exponential or other pattern. įirst of all: I agree with you - trendlines are not an exact science, but a help. I can send lots of stuff by e-mail of course. As a new user in TechNet, I am still suffering from some drawbacks - like the one that I cannot post links nor images.
0 kommentar(er)
