topic Re: How to predict an output on a table of inputs based on a Gaussian Process model ? in Discussions

How to predict an output on a table of inputs based on a Gaussian Process model ?

adroit — Wed, 28 Feb 2018 15:23:44 GMT

Hi,

I am a new user of JMP and I have done a design of experiment (Latin Hypercube) that gives me the values of an output (Y) depending on 3 inputs (X1, X2, X3).

I have then analyzed this (X, X2, X3, Y) table with a Gaussian process model.

I now would like to use this model to predict Y on a dedicated table of (X1, X2, X3) values.

Is this possible ?

In the model I can predict Y at one (X1, X2,X3) using the Prediction Profiler, but I would like to do the prediction on a big table of (X1, X2, X3).

Regards,

Arthur

Re: How to predict an output on a table of inputs based on a Gaussian Process model ?

cwillden — Wed, 28 Feb 2018 16:31:16 GMT

Hi @adroit,

Click the red-arrow next to "Gaussian Process Model" in the report and select "Save Prediction Formula." This will create a new column in your table with a formula that produces the predicted values from the values of X1, X2, and X3. You can actually just copy the formula over into a new column in the big table you're describing. If the columns have the exact same names for X1, X2, and X3 between the 2 tables then it should work immediately.