"unhide and unexclude" adds datapoints to multivariate analysis; can that be avoided?

[#1]

Hi,

I have a datatble with different categories of batteries that I want to analyses separately. So I have written a script that loops through the different types of batteries. It "hides and excludes" all batteries but the ones I'm interested in, then makes some plots. In the next iteration, it selects another type of batteries, "hides and excludes" the other ones and make some plots.

For the time being the plots I'm using are "Distribution" and "Multivariate" (analysis). The problem is now that when I'm finished with the first group of batteries and "unhide and unexclude" all data in the table to select the new group, the "unhided and unexcluded" batteries are added to the multivariate plots (the "Automatic Recalc" option is not selected). The "unhided and unexcluded" data are not added to the "Distribution" (i.e. histograms), only to the "multivariate analysis". "Unhiding and unexcluding" the data manually or via the script leads to the same result. Is there a way to prevent this from happening?

The effect can be seen in the attached file. "unhiding and unexcluding" the third datapoint will make it appear in the multivariate analysis but not in the distributions.

[#2]

This command affects the row states in the data table. They have a global effect through dynamic linking with any open platforms. You might use a Rows > Data Filter to select each group in turn. Alternatively, you could cast the grouping variable in the By role when you launch Distribution or Multivariate.

[#3]

Thanks Mark,
Using the "By" role would not work because the column I need to plot depend on the type of battery.

I will give the data filter a try.

[#4 · Accepted solution]

Two other options:

• Launch the platforms with a Where() clause to specify the subset of each report.  This will create a linked subset, so row-states will still remain linked with the data table, and you could use global or local filters in addition.  This is similar to what By() does, but you have more flexibility in specifying the subsets to be used.
• If you want to completely separate from the source table, you could create your own the subset tables as you are selecting the rows that you want to use in each set of reports.

[#5]

Using the data filter before lunching the platform does not give the desired result. The data are selected, but still all points are plotted.

I had thought about creating sub-tables but didn't want the end up with too many tables. Adding the Where() to the launch command of the platform seems to work nicely. I will continue in that direction.