topic Re: Reporting x value when y approaches zero in Discussions

Reporting x value when y approaches zero

thedellette — Fri, 09 Jun 2023 19:57:22 GMT

I have multiple data sets similar to the one shown below where I would like to review the y values (in the graph shown this is "H2O pickup") and create a report out when these values approach zero. I would like to be able to set a threshold around this at a value close to zero. In the picture I have shown -0.0005 for the y value and the resulting x solution reported would be around a cycle time of 1066. Please let me know if more data or information is needed to answer this. Thanks!

Re: Reporting x value when y approaches zero

pauldeen — Thu, 23 Sep 2021 08:15:57 GMT

Probably the easiest way is to load up the non-linear platform under analyze --> specialized modeling. Fit a good model and then use the inverse prediction for your target value to get you CycleTime.

Re: Reporting x value when y approaches zero

thedellette — Mon, 27 Sep 2021 12:54:38 GMT

Thank you! I can save the inverse prediction to the data table but I do not know how to save this formula to a script so that I can solve for a specific "H2O pickup" value and output a CycleTime value. Do you know where I might figure out how to do that?

Re: Reporting x value when y approaches zero

ian_jmp — Mon, 27 Sep 2021 13:57:22 GMT

Maybe something like this?

NamesDefaultToHere(1);

// Sample data
dt = Open("$SAMPLE_DATA/Nonlinear Examples/Negative Exponential.jmp");

// Fit a plausible model, and do inverse prediction for Y = 0.8 and 0.9
fc = dt << Fit Curve(Y( :Y ), X( :X ), Fit Logistic 2P(Custom Inverse Prediction( Response( {0.8, 0.9} ))));

// Get the inverse predictions into a table
dtp = Report(fc)[TableBox(4)] << makeDataTable;

Re: Reporting x value when y approaches zero

pauldeen — Wed, 29 Sep 2021 13:46:29 GMT

Or a slight modification to Ians method where the equation and predicted values end up in your original table:

NamesDefaultToHere(1);

// Sample data
dt = Open("$SAMPLE_DATA/Nonlinear Examples/Negative Exponential.jmp");

// Fit a plausible model, and do inverse prediction for Y = 0.8 and 0.9
fc = dt << Fit Curve(Y( :Y ), X( :X ), Fit Logistic 2P(Save Prediction Formula));

// Add desired row
dt << add row(1)
:X[nrows()] = 0.8;