Overlaying Data from Multiple Tables onto the Same Graph

[#1]

Hello,

I am a grad student who is new to JMP (I downloaded it a couple days ago) working on characterizing a new piece of equipment. I have a standard solution that I have added 20 things to at a known concentration. I then measure the concentration multiple times on the new equipment and I would like to make a graph that has box plots that show the spread of the detected concentrations (which I have done with graph builder) but then I would like to overlay points that show the true concentration on top of it (poorly drawn and shown in figure 2, the red points are the "true" values and the box plots show the detected values).

I currently have two tables. The first table has the data with the columns containing concentration data measured by the instrument. The second table has the same column headings but only has one row, which contains the true concentration of each component of the mixture. Even if I combine tables, I am unable to get the output that I am looking for.

Thanks for your help!

[#2 · Accepted solution]

JMP generally works with only one data table, known as the "current data table." Combine the two tables into one table. Create a new data column to indicate which original table a given row came from. You can use this new indicator in the Overlay role in Graph Builder.

[#3]

Hello, thanks for your suggestion.

That works mostly fine, except for the fact that it seems like that point is getting included in the box plot calculations. I want the box plots to only be based on the 20 detected values and not include the "true" value as a 21st point to calculate mean and quantiles from. Let me know if you need more clarification on what I'm trying to say.

Thank you!

[#4]

Actually, I think I got it figured out! I don't think it is including the 21st point, it just changes the view so that it seems like the box plots are changing but I think that their values are consistent.

Thank you so much!

[#5]

Here is my partial solution

Using the sample data table, Big Class, I use the Height column as the target column.  I add new rows, one row for each level of my X column, which im my example is the Age column.  These are your "True" values.  Next I add a character column that I called type and gave the value of "Raw Data" for the original data and "True Values" for the new rows

Then I opened Graph Builder, and dragged column Height to the Y axis drop area, and the column Age to the X axis drop area.

Right clicking on the graph I selected Add=>Box Plot

Right click on the graph and then select Points=>Jitter=>None

Then I dragged the column Type to the "Color" drop area

The raw data points need to be erased.  To do this, in the legend, right click on the blue dot marker for Raw Data and select  Color and then choose the White color box

Getting pretty close now.  Notice there is an errant blue data point in the above the Age 14 Box Plot.  This is an outlier point.  This can be hidden as were the normal data points by using the custom control dialog.

Right click on the graph and select "Customize" and then select "Outliers"

Click on the Line Color: box and select the White color box

I think this is the closest one can get to the graph you want.

It does have one error, the single True data point, is being calculated in the construction fo the box plot, but if you have a fair number of Raw Data values, the True value will make little difference.

[#6]

Thanks for taking the time to reply. I think that your solution will work at least for most of my data. There are a few analytes that it is important that the "true" value not be included in the box plot calculation since it is fairly distant from the measured values, but I think that using the previously suggested overlay solution will work for those.