more curve fits malfunctioned after working.

JohnFrench · Post by **JohnFrench** » Tue May 15, 2007 1:57 pm

Generate>More curve fits displays the fit equation in the legend along with a sample line. However the plot of the fit line does not appear on the graph. I have used the calculated coefficients to calculate x for a given y and the results seem correct. I just need the calculated line to be plotted. Is there some check box or something that I changed? This is the second set of data I have attempted to fit. The first set (very similar data) worked as expected; i.e. fit lines in addition to the data lines all appear on the graph. I am using the "dose response" fit. y=A+B/(1+10^(C+Dx)). Just as a test, I tried a second order polynomial fit and it worked displayed as expected.

Thanks,
John

Post by **DPlotAdmin** » Tue May 15, 2007 4:00 pm

Off the top of my head my best guess is the fit is lousy and is outside the extents of the graph, or perhaps inappropriate for whatever scaling you're using. Try 1) selecting Options>Extents/Intervals/Size. If "Specify extents" is checked, uncheck it. 2) Switch to linear X, linear Y from whatever scaling you're using.

If the fitted curve still doesn't show up please save your plot as a DPlot file and send it to me at your convenience.

JohnFrench · Post by **JohnFrench** » Thu May 17, 2007 4:25 pm

I now have it working properly. The only change I made was to remove one data series which had only one point. The rest of the data series have 13 points. Is it predictable behavior? In other words did that series with only one point cause the lack of display of the curve fit?

John

Post by **DPlotAdmin** » Thu May 17, 2007 4:47 pm

That would depend on the range in the data. If, for example, all of your series with 13 points were 0<X<1 and the lone point was at X=1000 or so, then it would have been drawn at the right extent of the graph (assuming symbols were used for that series; if not then it wouldn't be drawn at all) and the 13-point series would have been drawn as a vertical line at the left side, possibly obscured by the box around the graph. The same thing applies with Y: if the lone point had a Y value well outside the Y extents of the other series, then the graph might have appeared to be featureless.

Other than scale issues, though: no, there's nothing magical about a single point that would cause other curves to not be drawn.