I am using JMP to find a relationship between drought index values and yearly corn yields for a 30 year period. I have drought data for seven different indeces and each index ranges from -6 (severe drought) to +6 (extremely high precipitation).

The drought data contains monthly drought index values for March-September for every year from 1981 to 2011. The corn yield data contains one number (bushels/acre) for every year from 1981 to 2011. I built two side-by-side graphs in JMP, with one showing corn yield data over the 30 year period and the other showing drought data from one particular index over the same time period. Both are smooth curves and visually line up and seem to have a relationship.

My problem is that I have no idea how JMP is processing the drought data. There are eight separate drought index values per year, every year, for each drought index, and only one value per year for the corn yield data. I assumed that JMP was smoothing out the drought curve by taking an average of the eight index values for every year, then using that number in the graph construction but I haven't been able to find any literature that can tell me this for certain.

  • 1
    $\begingroup$ Can you post the graphs and/or give more specifics on how you made them (eg did you use the graph builder or were they generated with a different JMP module)? $\endgroup$
    – TLJ
    Oct 27 '13 at 23:39
  • $\begingroup$ !enter image description here Thank you for your answer. I simply dragged the variables I was interested in onto their appropriate areas in the graph builder in JMP 9. I did not use the Smoother element. Do you think the cubic spline explanation would be applied automatically? $\endgroup$
    – user32063
    Oct 29 '13 at 20:22

Just to make sure I understand you, I think you're plotting the index value by the year value, and there are several index values per year value. And you're in Graph Builder using the Smoother element. Here's a quick mock-up:

JMP Graph Builder smoother with duplicate X values

Actually in my mock-up, I also have the Points element turned on to emphasize the multiple Y values by X.

From the JMP 11 doc:

The smoother is a cubic spline with a default lambda of 0.05 and standardized X values. You can change the value of lambda using the slider. You can obtain the same spline in the Bivariate platform...

Cubic splines are technically only defined for data sets with unique X values. In case of duplicate Xs, JMP first takes the weighted mean of the corresponding Y values. Use the Freq drop zone in Graph Builder if you want to control the weighting, otherwise each Y is weighted equally.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.