A fellow consultant has advised that, when working on submissions for a regulatory agency, you should “model” your submission on previous ones (that is, leverage previous successful work to your benefit). In that spirit, I fully admit to “modeling” this blog post on those by Jeff Perkinson (10 Things You Don’t know about JMP and 10 More Things You Don’t Know About JMP).
With each release, the depth of JMP increases. Here are 10 things that you may not know that you can do with JMP. A few of these options are rather new (first appeared in JMP 12) while others are a bit more dated (JMP 9 and earlier).
Select one or more columns in the data table and then right-click in the column heading area on the table for a column menu and select New Formula Column. The options available will depend on the type of column selected (numeric or character) and the number of columns (single vs. multiple) selected. A new column with the formula selected will be added to your data table. This is a great way to concatenate character columns and to transform data (such as log transformations).
The default quantiles in the distribution platform are 0, 0.5, 2.5, 10, 25, 50, 75, 90, 97.5, 99.5, and 100%. These are easy to change. Under the red triangle menu on the variable of interest, select Display Options>Set Quantile Increment. This will pop up a dialog box in which to enter the desired increment (not broadcastable). Custom Quantiles, on the other hand, provides specific quantiles and their confidence intervals.
In large tables, grouping columns can help you organize your table and speed the entry of variables into dialogs (such as for fitting models). Highlight the columns you wish to group in the column panel, right-click for menu and select Group Columns. Once grouped, you can click on the grouped name to rename as appropriate.
The Effect Summary Report appears when you use any of the following model platforms:
For a single response, the p-values will match those in the Effects Tests table. For multiple responses, the Effect Summary is an overall summary reporting the minimum p-value across models for that effect. Adding or removing an effect applies to all models. Highlight an effect and click Remove to remove. Click the Add button to add effects. To add interactions or cross terms, use the Edit button. Thus, the Add and Edit button are similar except the Add is limited to adding main effects to the model.
While Graph Builder is awesome, don’t neglect the use of the Fit Y by X platform for all of the analysis capabilities found there. For example, with a categorical x (i.e., groups) and a continuous y (i.e., a measurement), the analysis of means (ANOM) can be used to compare group means. ANOM is a multiple comparison technique that compares the mean of each group to the overall mean with the results displayed in an easy-to-read chart. If a group mean falls beyond the decision limits, then it differs from the overall mean. Thus, you not only discover if the means differ but also which means differ, and the direction and magnitude of the difference. ANOM for Proportions is an option for categorical groups when the outcome is dichotomous.
For options to customize graphics, adjust an axis, or format a table, the options are often a right-click away. Position your cursor over a graph, axis, or table and right-click for options.
Make Data Table and Make Combined Data Table are right-click options for tables of results in platforms. The combined option will find all reports of the type selected and make one data table.
The Nonlinear platform has an option for comparing the parameter estimates of a set of curves. For instance, are the calibration curves from three lots the same or different? To do so, use a grouping column in the Nonlinear platform. Then once you have fit a model, the Compare Parameter Estimates option will be in under the red triangle for that model.
To quickly obtain numerical summaries of a group of columns, use the Columns Viewer from the column menu. Note that once you have your numerical summary, you can easily run distributions on selected columns. Remember that a histogram and/or box plot will give you a much better understanding of your data than a table of numerical summaries.
Platforms such as Distribution and Fit Y by X have an option to Arrange in Rows when you have multiple instances in a single window. This option allows you to select the number of reports to have in each row. This ability to arrange your results is helpful to improve the use of your screen space (there is also an Order by Goodness of Fit option in the modeling platforms).
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.