# Fitting data to plot using errors as weights [closed]

I have some enzyme kinetics data that I want to fit to the Michaelis-Menten curve in order to obtain the kinetic parameters of the enzyme.

Is it meaningful, statistically and functionally speaking, to use the error bars (standard deviation) of the measurements as weights in the NonlinearModelFit function, and are there any caveats that I need to take note of when using the function this way?

Needs["ErrorBarPlots"]
rateList =
Transpose[{{0, 0.1, 0.3, 0.5, 1, 3, 10}, {0, 0.0107, 0.0179,
0.02103, 0.0314, 0.0645, 0.0875}}];
errorList = {0.0015, 0.00043, 0.00071, 0.0043, 0.0051, 0.0050, 0.0071};
weightList = 1/errorList;
mmfit = NonlinearModelFit[rateList, a*x/(b + x), {a, b}, x,
Weights -> weightList]
Show[ErrorListPlot[
Table[{rateList[[i]], ErrorBar[errorList[[i]]]}, {i,
Length[rateList]}], AxesLabel -> {"[Substrate] (mM)", "Activity (Units)"},
PlotLabel -> "Michaelis-Menten Curve"],
Plot[{mmfit[x]}, {x, 0, 10}], Frame -> True]


Also, the AxesLabel property of the plot does not seem to be displaying the axes' names in the resulting plot. What am I doing wrong in this case that is preventing the axes from being correctly displayed? • The standard practice, as noted here, is to weight by variance; so, weightList = 1/errorList^2 seems more appropriate. May 21, 2015 at 19:19
• You have set Frame to True so you need FrameLabels rather than AxesLabels. May 21, 2015 at 19:19
• @Guesswhoitis. After changing my weights to use the square of the standard deviations, the curve is now cut off. I presume this is again due to some automated plot shenanigans near an asymptote, but I can't find the correct option to use in the Plot documentation to fix this. May 21, 2015 at 20:02
• ... or, possibly more easily, just add PlotRange->Full to the Plot` function. May 21, 2015 at 20:10
• You might want to see this for your future reference. May 21, 2015 at 20:20