Subscribe Bookmark RSS Feed

Can a platform be modified after it has run

rich_greves

Community Trekker

Joined:

Sep 4, 2014

I am new to the JMP community, and am trying to learn how to use JMP.  I am analyzing monitoring data.  We have 54 compounds, 40 sites and either 1 hour, 3 hour or 24 hour data.  If I want to run a platform (distribution), I set it up to run for one of the compounds (benzene - 24 hour) and it works fine.  If I wanted to run the same analysis for toluene - 24 hour, is there a way to modify the inputs so that it now looks at the toluene column, or do I have to create a new file for all of the combinations (54 compounds * 40 sites * 3 time parameters?)

1 ACCEPTED SOLUTION

Accepted Solutions
Solution

Yes, the problem is the separate data files. From an analysis perspective, it is expected that similar items will appear in the same table. As an outrageous example, if I use your table to make a scatterplot of benzene vs. toluene, I could open up a second data table that has information from my favorite football team. Dragging a variable from the new table onto the existing scatterplot would not make much sense.

That being said, if your data files are all structured the same, it would be easy to use the Tables > Concatenate command to combine several separate files into one file. Also, for any analysis you can save a script for that analysis and run the script against any other data table. You have to do some minor changes to script to make it generic for the variable to use in the analysis (or have the same variable name in every data table). Ultimately, scripting would likely be the best way to handle your problem with the separate data files.

Dan Obermiller
3 REPLIES
Dan_Obermiller

Joined:

Apr 3, 2013

The answer to this question will depend on how your data table is structured. What you need is a single file with all of the data. That gives JMP the most flexibility and allows JMP to show its true power! A file (I will use capital letters to indicate a column name so it is easier to refer to) with a column for COMPOUND, another column of SITE, a column with HOUR, and finally, a column with the MEASUREMENT would work nicely. Alternatively, you could have three columns for the measurements: a 1 HOUR, 3 HOUR, and 24 HOUR which would take the place of the HOUR and MEASUREMENT columns.

With the data structured this way, you could create your first distribution. To reproduce this analysis for a new variable, you have several options:

You can go to the red pop-up menu and relaunch the analysis to choose a new variable.

You can go to the red pop-up menu and choose the Column Switcher which allows you to choose the other columns you may wish to examine.

You can even drag and drop variables into the analysis as shown with this video: JMP: How to Drag and Drop Variables to Update an Analysis - YouTube

To change the COMPOUND or SITE, you would set those items up using either the Data Filter (found under the Rows menu) or the Local Data Filter (under the red pop-up menu from the distribution analysis). The Filters allow you to quickly subset the data by the various compounds or sites and the graph would automatically be updated for the levels you choose.

I hope this helps, but if not, please provide some more details of your data structure.

Dan Obermiller
rich_greves

Community Trekker

Joined:

Sep 4, 2014

I guess my data is structured poorly because I don't see any of the options that you mention.   My data is as such.  A different JMP file for each site/time combination.  In other words there is a separate JMP file for site1 -1-hour, site1 -3 hour site1 - 24 hour ...  I have a column for each compound, and each row is a different date.  So an abbreviated example of my table is:

Date                               Year    Month      Day   Benzene              Toluene           Ethylbenzene 

19930701Average            1993    07           01      0.112380952        0.112346939    0.075535714

19930702 Average           1993    07           02      0.198333333        0.396607143    0.05578125

19930705 Average           1993    07           05      0 0 0.0775           0.465               0.5425

19930706 Average           1993    07           06      0.215138889        0.303571429    0.0771875

19930707 Average           1993    07           07      0.239375             0.42672619      0.0740625

19930708 Average           1993    07           08      0.210681818        0.430454545    0.086534091

19930709 Average           1993    07           09      0.305396825        0.573333333    0.094821429

19930710 Average           1993    07           10      0.371594203        0.709937888    0.105543478

19930715 Average           1993    07           15      0.091666667        0.137142857    0.04875

19930716 Average           1993    07           16      0.074333333        0.116857143    0.07375

Just one of my sites has 2853 rows and 84 columns, so combining all of my data into one table doesn't seem feasible to me

Solution

Yes, the problem is the separate data files. From an analysis perspective, it is expected that similar items will appear in the same table. As an outrageous example, if I use your table to make a scatterplot of benzene vs. toluene, I could open up a second data table that has information from my favorite football team. Dragging a variable from the new table onto the existing scatterplot would not make much sense.

That being said, if your data files are all structured the same, it would be easy to use the Tables > Concatenate command to combine several separate files into one file. Also, for any analysis you can save a script for that analysis and run the script against any other data table. You have to do some minor changes to script to make it generic for the variable to use in the analysis (or have the same variable name in every data table). Ultimately, scripting would likely be the best way to handle your problem with the separate data files.

Dan Obermiller