This example constructs a response surface model using the Odor Control Original.jmp file in the Sample Data folder. The objective is to find the range of temperature (temp), gas-liquid ratio (gl ratio), and height (ht) values that minimize the odor of a chemical production process.
 1 Select Help > Sample Data Library and open Odor Control Original.jmp.
 2 Select Analyze > Fit Model.
 3 Select odor and click Y.
 4 Select temp, gl ratio, and ht, and click Macros > Response Surface.
This Macro adds terms up to degree two to the model (Fit Model Dialog for Response Surface Analysis). The main effects appear in the Construct Model Effects list with a &RS suffix. As a result, the analysis results will include a Response Surface report.
 5 Click Run.
Fit Model Dialog for Response Surface Analysis
The Parameter Estimates table shows estimates of the model parameters (Parameter Estimates Table). Two of the quadratic effects, gl ratio*gl ratio and temp*temp, as well as the main effect of ht, are significant at the 0.05 level.
Parameter Estimates Table
 6 From the report’s red triangle menu, select Save Columns > Prediction Formula.
This inserts a column into the data table called Pred Formula odor. This column contains the prediction formula defined by the coefficients shown in the Parameter Estimates table.
A response surface report is provided (Response Surface Report). This report shows a table showing the second-order model coefficients in matrix form. The Solution report shows the critical values for the main effects. These are the values where a maximum, a minimum, or a saddle point occur. The Solution report indicates which of these occurs at the critical point. In this example, the response surface achieves a minimum at the critical value.
Response Surface Report
The Response Surface report also contains the Canonical Curvature subreport (Basic Reports for Response Surface Model). This report shows the eigenstructure of the matrix of second-order parameter estimates. The eigenstructure is useful for identifying the shape and orientation of the curvature. The eigenvalues, given in the first row of the Canonical Curvature table, are negative if the response surface curves down from a maximum. The eigenvalues are positive if the surface curves up from a minimum. If the eigenvalues are mixed, the surface is saddle shaped, curving up in one direction and down in another direction.
Basic Reports for Response Surface Model
Prediction Profiler
 7 From the Prediction Profiler’s red triangle menu, select Desirability Functions.
In the top row for odor, to the far right, a cell that plots the desirability function is added. A row of cells showing desirability traces is added beneath the row for odor.
 8 In the upper right cell, hold the Control key and click.
 9 In the drop-down list, select Minimize.
 10 Click OK.
 11 From the Prediction Profiler’s red triangle menu, select Maximize Desirability.
Prediction Profiler with Desirability Function Optimized