Hide and Exclude individual cell

[craig_king]

In the data table I will have 20 columns and 50 rows. Column 12 row 36 has data I have collected that is suspect. I would like to exclude the individual data point without excluding the entire row or column. Perhaps treating the cell as a blank cell without having to delete the data point.

It would be nice if there was a graph option during the right click on a data point to exclude row or exclude individual data point. Also a right click on the data table cell option to exclude cell.

It would also be nice to be able to include a comment for the individual cell that is attached to the cell by right clicking the mouse.

[mia_stephens]

One currently available option is to specify suspect values for a column as missing by using the Missing Value Codes column property. Interesting idea to add this as a right-click option from a graph or cell. 

[hogi]

---

@craig_king wrote:

It would be nice if there was a graph option during the right click on a data point to exclude row or exclude individual data point.

---

In https://community.jmp.com/t5/Abstracts/Stay-in-the-Flow-With-Custom-Scripts-and-Toolbars-2022-US-30M... @ 12-14min, 
@shampton82 shows a way to "Nuke" data points.

Concerning:

Perhaps treating the cell as a blank cell without having to delete the data point.

---

one has to decide which cell - the column on the x or on the y axis - or maybe the one that is used as page?
@shampton82 uses a nice trick:

Names Default To Here(1);

dt=Current Data Table();
cnames = Current Report()[listboxbox( 1 )] << get selected;
rowindex = dt << Get Selected Rows;
dt[rowindex, cnames] = .;

The fist ListBox in the report is used to specify the columns. It's quite universal, it can be a Column Switcher, a Local Data Filter or the standard Column Menu of Graph Builder:

The disadvantage of this approach:
the cells are not just treated as blank - the value in the cells are deleted. Un-Hide is not possible.

[hogi]

As an extension of the Nuke approach:

Before deleting the values, one could store the values in an auxiliary column.
Maybe: an auxiliary column with vector type - with pairs of source coulmn and the stored value.
This way, one could use the same auxiliary column to store all nuked values -- and to restore them if this is needed.

[hogi]

The possibility to exclude individual cell comes with a huge danger: this information is "hidden", the user has to remember it!
Perhaps it would be better to provide it as an optional feature that can be enabled if needed, in order to protect all "innocent" users?

A very conservative approach:
The information about nuked cells is stored in a column Nuked with property Multiple Response:

To nuke a cell in column XYZ and row N , one has to add "X_XYZ" to Nuked in row N.
In the examples, for row 478, the values of columns PNP5 and NPN3 are nuked.

This information can be used in a report - by using the column Nuked in the data filter - with the option "Match none":

A user who wants to nuke outliers in columns PNP5 and VPM3 can select the entries in the data filter:

One can automate this selection via a state handler of a column switcher.