How to filter interactively for box plots?

[k-wizzz]

Hi fellow KNIMErs,

This is bugging me for quite some while: I created a Component to enable a user filtering for certain criteria and then displaying the filtered data/numbers as a (conditional) box plot.

It looks like the available box plot nodes do not have an Interactivity option as others (pie chart, scatter plot, etc). Filtering/selecting with the component does not update the box plot; it always shows the results unfiltered.

Background story: I want to see the salary distribution based on org units, job category, and salary. (Input file has org unit, job category, salary; one line per employee.)
Then I want to ask questions like “Show me distribution of salary for given job categories amongst selected org units”.
The box plot would then display boxes for each selected org unit.

Maybe there is some trick I’m not aware of? Or another node that I didn’t find?

Thanks for any pointers/hints!

[mlauber71]

@k-wizzz maybe you can take a look at this article which contains an example with a boxplot and a component for interaction:

Create an Interactive Dashboard with KNIME Components and Python

Interactive dashboard where bar chart and table consume same filter

@cdruf_iu here is an example of a dashboard where you could select a string variable and a numeric variable which then will be displayed in a Boxplot. Also you could select one category from the string variable and sort the rows by the numeric variable descending. Also you can download the graphic as a PNG file with an individual name. [image] You might be able to adapt that to your needs [image]

[k-wizzz]

Cool. Thanks!
So, the “only” magic that’s inside is the Refresh Button Widget, meaning that box plot has no real interactivity as other Javascript graphical nodes that share their selections/filters within a component automatically.
Instead, I do the filtering as usual (adding some interactive filtering nodes) and use the Refresh Button Widget to tell KNIME to recalculate things, right?