Why can't JMP allow Special Fit lines to be extended beyond data set?

[nocaltim]

     This is a request to add back a feature/capability that existed before JMP 13.

After making an XY plot, it is common to have JMP apply a best fit line.  However, JMP limits the extent range to be not much beyond the X-values of the data set.  This is to avoid extrapolation errors and it is, arguably, a good thing.  

     However, I frequently plot XY plots then want to apply a forced line.  For example, I have two columns of data representing a single measurement reading made by two different pieces of test equipment.  Notwithstanding R&R, each of the readings should be the same value regardless which piece of test equipment collected the reading.  Therefore, I want to apply a "special fit" -- a  forced 1:1 line fit (slope=1, intercept = 0).  Please note that I *know* that this 1:1 relationship should exist far beyond  the range of the data plotted.  The problem is that JMP 13 refuses to extend the 1:1 ("special fit") line beyond the data.  JMP thinks it is protecting me from using best-fits and extrapolating beyond safe ranges, but in this case, this is not a "best-fit", it is a forced-fit and, in this case, I know with 100% certainty that the forced-fit is valid out to very large range extents.  I do not want to have do special scripting to create a graphical equivalent of this request.

     REQUEST:  In XY plots, JMP should allow forced-fits to extend the full range of the displayed axis limits, even if those axis limits are far beyond the plotted data.

[oxap]

Strongly agree with the request. There are a number of techniques in electrical engineering that require fitting a line to data, and using the x or y intercept to extract parameters. For example, TLM (transfer length method or transmission line model) measurement: measure the resistance of a resistor as a function of length. By extrapolating to zero length, you find the portion of the resistance contributed by the contacts. Graphically, it is nice to see the extrapolation to the axis limit. Yes, things can be extrapolated carelessly, but this should be up to the user to decide. This is one of many types of models that are perfectly valid extrapolations. Unfortunately, being unable to do this simply in JMP graph builder has led me to transition this portion of my work to python. 

[andreacoombs1]

Graphs in JMP can be customized to include a line that is a function of x.  

- Right-click on the white space of the graph

- Select Customize...

- Use the + Button to add a script

- Add a Y Function()

JMP has Templates and Samples you can use.  Under the Samples, I added the Sine Wave which you can see extends beyond the range of the data.  You can edit the Y Function() to include your function of interest.

[Ryan_Gilmore]

We are archiving this request. If this is still important please comment with additional details and we will reopen. Thank you!

[oxap]

I don't know why I got an email saying this post was just updated. However, since I'm here, and I hadn't seen the original response, I'll add another comment - the proposed solution is clunky and inconvenient. Each time I do this very basic fit, I have to add a function and type in the fit parameters? What if I have 5 lines fit on a graph? 

[nocaltim]

Agreed -- the proposed "solution" to this problem does not provide a "value add" to the issue that was raised.  I appreciate the effort to put forth a solution, but this "solution" is too clunky and unnecessary to address what is a very simple issue/problem.   At this juncture, I've come to the conclusion that SAS doesn't want to support this request -- not that they can't, just that they won't.  Would sure be nice if SAS would  be more forthright and say so.  

[andreacoombs1]

Just got an email today as well.  There is another to approach this... You could also use the Fit Y by X platform under the Analyze menu.  Under the Red Triangle there is an option called "Fit Special..." where you can constrain your slope and intercept.  The bivariate fit in Fit Y by X will extend any fit lines beyond the data.  This graph has the similple linear regression line (green) and a line with slope and intercept contrained to 0 and 1, respectively (red).