Use of filters and saving insights for reuse

Janet Wittenberg

By using Insight Designer to explore your data, you’ll discover new patterns in your data.  You can export the results to Excel with a formatted tabular view of the data from the visualization.  Or you might just grab a screenshot to help you pass on the information another way.  Use the filters at the top of the Insight designer to simply and quickly adjust the data included in your visualization. These filters get added automatically when you select attributes; you can also drag more attributes on to the filter area to restrict your data further. 

Typically, you’ll create insights to share out with other people by placing them on a dashboard, so it’s important to understand how the layers of filter options work. 

 

Besides picking the type of visualization for an insight, and what you are measuring, the next most important decision is how to filter.  By the time an insight gets on a dashboard, there are actually 5 different places where you can filter the result. 

(from least controlling to most controlling) 

1. Filters at top of insight designer “insight page filters” 

2. Filters on dashboard “dashboard page filters” 

3. Filters in Properties on insight placed on a dashboard  
   
   - choose to ignore page filter “disable dashboard page filter” 
   
   - define additional insight filters “insight properties filters” 

4. Filters on a measure in your insight “measure attribute filters” 

Ultimately your insight is analyzing a dataset.  Using an analogy in Excel, this would be like the rows you choose to include using the filter feature.  All of the filter types except measure attribute filters affect which rows are included in the dataset analyzed by the Insight.  When working in Insight designer, the insight page filters limit the data set for the purpose of viewing within Insight designer or exporting your data.  However, when an insight is placed on a dashboard, these insight page filters will be superseded by a dashboard page filter using the same attribute.  If the dashboard page filter is set to “All”, your insight will render using rows with All (any) values for that attribute, negating your insight page filter.    

If the dashboard page does not include a filter for the attribute you have selected in the insight page filters, your insight will render as you expect.  Or, on a specific dashboard with a dashboard page filter that overrides your Insight page filter, you can still use Insight Properties to disable the dashboard page filter for this insight as if the dashboard page filter did not exist.  

In the Dashboard Builder, if the insight is placed on a dashboard and the Insight Properties are used to add an insight properties filter with the same attribute, the selected values will ultimately control which attribute values define the rows analyzed by your insight, superseding the dashboard page filter and the insight page filters you originally specified. 

Keep in mind that once your insight is saved from Insight Designer and available to be placed on any dashboard, you cannot control whether a future dashboard page will use a page filter or Insight Properties that can change your insight results.  A dashboard builder can modify your insight on any page using insight properties filters, and control whether a dashboard viewer can change vailable dashboard page filters to modify what your insight shows on a dashboard. 

Using measure attribute filters on a measure limits the data that will be counted in that measure no matter what.  When this kind of insight is placed on a dashboard, neither the dashboard page filters nor the insight properties filters will expand the rows included in that measure. The dashboard page filters or insight properties filters can reduce the rows measured to a subset of the attribute values originally built into the measure attribute filter. 


The first step to building and saving your insights is to determine what style of dashboard page it is intended to support.  Some dashboards are very specific to tell a particular story to the viewer without exposing any filters.  Some dashboards may be expose lots of dashboard page filters to allow the viewer to slice and dice and see the results of different filter combinations.  When you save an insight, use a naming convention that will help identify its purpose. 

Let's use a specific dashboard example: 

How you plan to name your insight should give you an indication of how to design the filters. 

For example, extracurricular involvement by gender. 

My definition of extracurricular involvement is “section enrollments for Athletics and Activities”.  I will choose to use Offering type as a measure attribute filter on section enrollments so that I am sure the integrity of my insight name will remain ‘extracurricular involvement’. 

Conversely, if I create “Group enrollment by Gender”, then this insight could be used along with a dashboard page filters or insight properties filter to create different visualizations for each of Academics (which courses students of different genders choose) versus Athletics (team participation), or Activities (clubs, etc).  This insight gives the dashboard builder and/or the dashboard viewer more flexibility in reusing this insight to analyze overall enrollment of various types at the school. 

Other ways a saved insight can be reused for more flexibility: 

The component ‘by gender’ likely drives the shape of the visualization you choose based on what looks good and the scale of the number of values (a short list of 2 or 3 values for gender for example).  However, when an Insight is placed on a page, a dashboard builder can use the Insight properties to override the attribute used as a row, segment or trend in your insight.  For example, the attribute Gender can be swapped with 'Boarding or Day', or 'Is child of faculty' on a line graph or pie chart.  Consider that swapping to something like ‘Zip code’ or ‘Section’ might not work because there will be a larger number of values or segments, which probably requires a different visualization like a bar chart or tabular view. 

This insight could be named generically as “Group enrollments by demographic” reminding you that several different demographic attributes could be used.