Model for air flow and dust pick upHi,
I have below data
Air_flow Dust_pickup
18.1 87,3
18,9 86,8
18,8 86,8
16,5 84,7
I would like to make a model which can explain ther realationship between Air_flow and Dust_pickup and then use this model to estimate the Air_flow required to get a Dust_pickup of 83%.
Which tool is the best to make this model and predict the value ?
/Tahir
Re: Model for air flow and dust pick upNo, you do not need JMP Pro to use the Prediction Profiler.
Sorry for the lack of details. I thought that you knew JMP but were asking about modeling.
Click the red triangle next to Fit Least Squares and select Factor Profiling > Profiler. Now click the red triangle next to Prediction Profiler and select Optimality and Desirability > Desirability Function. Control-click on the new Desirability profile on the right side and change the default goal Maximize > Match Target. (Now leave the column of desirability values alone.) Change the low, middle, and high to something that is reasonable for your case. I used (82.75, 83, 83.25). Now click the red triangle next to the Prediction Profiler again and select Optimality and Desirability > Maximize Desirability.
Re: Model for air flow and dust pick upThanks for the quick reply but where did you selected the Prediction profiler? I have JMP 13.1 Is it included in this version or it requries JMP Pro?
Re: Model for air flow and dust pick upThe data are deficient in sample size and range of both response and predictor. I just want to set your expectations at a realistic level.
I entered the four observations and explored the data with Graph Builder. It appears to be an inverted parabola, assuming normal errors and no outliers. I used Analyze > Fit Model to estimate parameters for a second order polynomial model in Air Flow. Here are the results.

Then I used the Prediction Profiler to find the Air Flow level that is predicted to yield Dust Pick Up of 83.

Please note the extremely wide confidence interval on this prediction due to the deficiencies mentioned above. This predictions is a large extrapolation of the model beyond the observed predictor range.