We're excited to announce that we're launching the second installment of Dataiku Product Days Register Now

Filter to show only records changed by specific Visual Prepare Recipes Steps

User Story:

As an analyst who wants to better understand the impact of changes when creating steps using visual recipes,  I'd like to be able to filter rows, of a sample, to see just rows that are changed by a particular visual step or a set of steps.  This would allow for better observability of the intended, and potentially un-intended impact of steps applied to a dataset,  This would also speed analyst workflow increasing productivity, and may ultimately lead to cleaner data.

COS:

  • Once a particular row filter is set, by default, this would be sticky to those particular rows, even if steps are changed that might make any particular row/rows fall out of the original filter set.
  • This filter should be easy to turn on and off like current column-specific row-based filters. 
  • Resetting the filter on the same steps should take fewer mouse clicks than setting up the original filter. (It would be great if there was a way to turn off the stickiness so that changes to the original steps changed the filter.  So one button that says refresh this filter. Or even auto-update step filter.)

Notes:

  • This might be implemented as a new filter type accessible under the Step Action Drop Down.
    • One would choose one or more steps with the little checkboxes
    • One would open the Action Menu above the steps and choose filter rows modified by selected steps.
    • This would put a filter "pill" above the dataset in the usual way that columns filters are seen today.
    • One could turn the filter on and off however once set the rows chosen would stay the same until the sample is changed.  Or you turn on auto-update Step Filters.
  • Today to do something like this.  You have to change your sample to show rows that are likely to be changed by your steps.  You often miss, unintended consequences because your sample filter does not actually include all rows actually impacted by the steps you are working on.
  • Today if you don't use the sample filter method, you might have to scroll down several pages of records to find records impacted by particular steps.  If this is several screens down you have to wait for rows to be displayed.  If you change a step you are automatically sent to the top of your sample and have to scroll down again. (ugh....)
1 Comment
AshleyW
Dataiker
Dataiker
Status changed to: Acknowledged

Thanks for your idea, @tgb417. Your idea meets the criteria for submission, we'll reach out should we require more information.

If you’re reading this post and think this would be a great capability to add to DSS, be sure to kudos the original post! Feel free to leave a comment in the discussion about how this capability would help you or your team.

Take care,
Ashley