While using predicting screening to identify the contribution of each individual parameter among the data each time i run the program on the data different portion and contribution is observed.
That is correct. Becasue the predictor screening uses boostrap forests to generate the contributions, each time you run the procdure you may see different resutls. The larger the number of predictors, the more variblity you will see in the results. I often run the procedure 5 times and take the top contributors from each round and compare or feed them forward to the next step of my analysis.
For more info on the platform here is the overview help page:
in JMP 14, you can set the number of trees, more trees result in more stable results (but they take a little longer to run).
more data helps too.
There is a text box on the diaglog window (bottom left) that allows you to set the number of trees. The default is 100.