JMP is all about data exploration, and a big part of exploring is not knowing the questions up front. JMP 11 adds a feature called “transform columns” that allows you to create a temporary computed column within almost any column list in JMP. That means you can create a new column right in Graph Builder or Tabulate or even the Local Data Filter without having to go back to the data table and interrupt the flow of exploration.

In the JMP Discovery 2013 keynote, I showed several transformations using data from the SAS solar farm. The data table's date/time data is at 15-minute intervals, but I wanted to look at the data by month. So I right-clicked on the Date Time column and applied a Month Year transformation.

And I wanted to look at the efficiency of the panels, so I selected Power and Irradiance (the amount of solar energy at the site), and applied the Ratio transformation to get an efficiency measure.

With those operations, I get two new columns to use in the graph: "Month Year" and "Power/Irradiance."

Using box plots, I can see there is a fairly consistent monthly median efficiency and some extreme outliers, which turned out to be bad irradiance values.

At this point, the columns exist only in Graph Builder and not in the data table, but I can right-click on the transform columns option and choose “Add to Data Table” if I want to keep them as real formula columns.

As the drop-down menus in the above pictures suggest, there are several categories of transformations to apply and many transformation functions within each category. Most are straightforward mathematical transformations, such as Ratio, Log, and Square Root. The Row submenu, however, contains some row-wise transformations like row number, cumulative sum and an exponentially weighted moving average (EWMA).

For example, the JMP sample data file Nile.jmp contains only one column, which is a series of historical flow levels for the Nile River. Using transform columns, we can explore the data over time in Graph Builder by adding columns for row number and the moving average.

You can learn more about this and other JMP 11 features by watching the on-demand Mastering JMP webcasts.

You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.