Guidelines for Custom Dataset Creation in Digital.ai Intelligence

A custom dataset is a blank report that contains metrics and attributes that can be used as a starting point for building a Dynamic or Dossier Dashboard. In certain use cases, it might not be possible to find all necessary data for a given dashboard within the existing cubes and a custom report is necessary.  

Note: Please create custom datasets only when necessary. If you are unsure if a custom dataset is required please consult your CSM.

When it has been determined that a custom dataset is necessary, please follow the guidelines below for best results:

  • First, check for similar datasets to avoid replications
  • Use only certified and necessary attributes and metrics for your analysis
  • After creating the report check the SQL for cross joins. If sub-optimal joins are present the system may not execute the report or return an error if it detects a bad query
  • Verify data against the source
  • Check the size of the report. Custom datasets should be limited to a few thousand rows. If your custom report is larger please consider using an existing cube
  • Once the dataset is certified please work with your CSM to migrate it into an iCube

To check the SQL of a report

  1. Execute the report.  While the report is loading click on the option "Show Report Details":
    1_-_Processing.png
  2. Under Report Details you will find the SQL statement that contains any bad joins, specifically "Cross Joins."  Cross joins can be caused by pulling in only attributes without a metric, neglecting to filter data across tables in any way.  Evidence in the SQL statement that a cross join is present can be repeated metrics across rows. Please see the screenshot in below to observe a cross join:
    2_-_Cross_Join.PNG
  3. An alternative way to view the SQL is to execute the report completely and then navigate to the "Tools" drop-down menu
  4. Select Report Details Page
    3_-_Report_Details.png
  5. Once on the details page, you will see a box labeled "SQL statement."
  6. Check the SQL for any bad joins
    4_-_SQL_Statement.png

If a bad join is observed or an error is returned while executing the report please contact support@numerify.com to continue troubleshooting.

 

Related Article: Pros and Cons of Using an iCube vs. a Custom Dataset in Digital.ai Intelligence

Was this article helpful?
0 out of 0 found this helpful

Comments

0 comments

Please sign in to leave a comment.