Skip to main content

More filtering options for Path Analyzer - Filter by module

Related products: PX Analytics

I would love to be able to filter the path analyzer report by module. So for example, if I wanted to examine one area of the product (ex. the process for creating a chart), I would like to filter by the associated module and see all the clicks within that module. This is useful because it would eliminate clicks that were unrelated to that process (ex. editing your profile picture). 

 

Another useful feature could be the ability to minimize “Other” in path analyzer so that we could see the features more clearly. 

New IdeaCurrent Behaviour

Hi @akarabatsos 
To tackle the others part of Path Analyzer, we can leverage the exclude with labels functionality. A label can be quickly created from the path analyzer to include the features already being shown on the path analyze report. Once this created label is excluded, this will enable more features grouped in Others to show up.
Will consider the module option in addition to the existing feature and engagement.


@Chandu 

While the idea of using labels for features or modules in Path Analyzer seems helpful in theory, it becomes impractical when dealing with over 1,500 features. The challenge is that it’s impossible to know ahead of time which labels will be useful for future analyses. Each analysis might require a different combination of features depending on the hypothesis being tested. As such, pre-defining labels becomes inefficient and can even limit the flexibility of the analysis.

A single label might work for one use case, but in another scenario, you’d need an entirely different set of features grouped under a different label. Since the value of the analysis depends on the ability to adapt based on different paths and hypotheses, this rigid structure prevents that adaptability.

This issue has already been highlighted in another topic on this forum:

Instead of focusing on labels, a more dynamic and flexible solution for feature selection is needed—one that doesn’t require pre-labeling or excluding hundreds of irrelevant features manually.