topic Re: Adding line to split data on graph builder in Discussions

Adding line to split data on graph builder

tarkan_bih — Sun, 11 Jun 2023 11:08:18 GMT

Hi,

I have the following data set below where red data points are some fails that I would like to screen out. Right now the way I am doing this is roughly calculating a trend line (see image below) and finding its equation. Then I create a new column and use JMP formula to basically check if data point is below line and if it is it updates the new column with a 1 and if it is above it marks new column as a 0. This is a pretty manual process and I was wondering if there was a way you could draw a line in graph builder the would autogenerate its equation and select all units below the line and if possible also create a new column?

Re: Adding line to split data on graph builder

Thierry_S — Sat, 21 Nov 2020 02:23:20 GMT

Hi,
Is this a typical distribution of your "Pass" and "Fail" data? If it is, it might be quite difficult to identify a rational process to separate the two populations because these broadly overlap (i.e. no apparent trend in distribution above and below your "decision" line). If it is not representative, could you provide a more contrasted example that could guide a possible approach?
Best,
TS

Re: Adding line to split data on graph builder

txnelson — Sat, 21 Nov 2020 02:38:10 GMT

Here is how I would do it

Create a new binning column and specify the following formula
```
If( Selected( Row State( Row() ) ),
	1,
	0
)
```
Go to your graph and then select from the tool bar, the lasso tool
Use the lasso selection tool to capture all of the data points you want declare as bad data points
Note that all of the rows in the data table that are associated with the selected point, are selected and the value in the new bin column has been set to 1.
Now right click on the Header for the bin column and select Col Info
In the Col Info window, click on the Remove button to delete the formula, thus turning the bin values into static values

Re: Adding line to split data on graph builder

Mark_Bailey — Sun, 22 Nov 2020 15:09:59 GMT

BTW, since Selected() is a Boolean function, the If() function is superfluous. This expression is all that you need for the column formula:

Selected( Row State( Row() ) )

Re: Adding line to split data on graph builder

ZF — Tue, 24 Nov 2020 00:05:51 GMT

I created a column "Line" with a line function, created another column "Bin":

the graph shown below:

Re: Adding line to split data on graph builder

tarkan_bih — Tue, 24 Nov 2020 00:50:54 GMT

How did you plot the line to split the points afterwards? I am able to get the overlay colors based on the new bin but just not the dashed line you have in graph builder.

Re: Adding line to split data on graph builder

txnelson — Tue, 24 Nov 2020 02:07:32 GMT

I did not draw a slashed line.....I thought you were just using that for selection, not display.

Re: Adding line to split data on graph builder

ZF — Tue, 24 Nov 2020 03:27:02 GMT

To plot the line, put both columns "weight" and "Line" on the Y axis. Then select both Scatter plot and fit line plot from the tool bar, uncheck Y Line under the "Point" and Y weight under "Line of Fit" on the left option bars. Please see screen shot below. thanks. ZF

Re: Adding line to split data on graph builder

tarkan_bih — Tue, 24 Nov 2020 17:53:54 GMT

Sorry for confusion, I wanted a line for both selection and display.

Re: Adding line to split data on graph builder

tarkan_bih — Tue, 24 Nov 2020 18:05:02 GMT

I was not aware of this functionality in Graph Builder thanks! I was able to add the line and simplify my workflow for this