Experience with organizational data type / featured tables

[IMett]

Hi everyone,

We are currently testing the "orgnizational data type" feature. (--> https://docs.microsoft.com/en-us/power-bi/collaborate-share/service-excel-featured-tables) 

All in all, the feature works well for us, and the business units are quite stunning about the possibilities.

However,  we noticed that the automatic recognition rate is lower than expected for several reasons:

- my feature table has the key column "employee" which is a character fields containing [prename surname] (e.g. "John Doe". When I try to categorize an employee entering "John Doe", Excel proposes all Johns in the organization, with John Doe often (but not always) in the first place.

- some values can occur in several columns. In this case, Excel doesn't know which column to choose, and often produces rather weird proposals.

So I am looking for best practices (if there are any) how to build the featured table, so that the recognition will be as good as possible. What I would like to do is:

- tell Excel to automatically categorize a cell as the best match, even if the confidence is below 100%

- tell Excel in which columns to look for a match with preference

Does anybody has any experience with those issues? I didn't find any ideas about that.

[kathleeng23]

I've had this same issue in that when we search for an employee, it also brings up everyone that reports to that employee.  It's a cool feature but would be nice if it sorted the results more accurately.

[collinq]

Hi @IMett ,

I am wondering about using fuzzy match to build your data so that you don't have to use 100% confidence?

[IMett]

Hi @collinq ,

yes, that's also one of the things I thought about. But how could I use this in the featured tables scenario?

[collinq]

Hey @IMett ,

This link has some of the hints behind how it works - maybe modify the Feature table to get better results?  Perhaps things like not using numbers ?  Set featured tables in Power BI Desktop - Power BI | Microsoft Docs

[IMett]

Hi @v-henryk-mstf 

any further ideas about this topic?

[v-henryk-mstf]

Hi @IMett ,

According to your description, when a match is found in the column of a table in Excel with a preference match. So what is your specific business scenario in powerbi? Can you provide a screenshot to describe it so that I can answer you as soon as possible?

Looking forward to your reply.

Best Regards,
Henry

[IMett]

Hi @v-henryk-mstf 

Thanks for your reply.
I had to create a fictional dataset, because the original one contains sensitive data.

This is my featured table (opened in Power BI Desktop). I registered it with "employeeId" as the key and "employee" as the display column.

Now this is what I encounter in Excel (GIF is probably the easiest way to demnostrate)

As you can see, when entering "John Doe", I get 5 possible results, with the first row from the master table is only in the 3rd position. The same occurs also for other items.

In my use case, I would like to use the featured table in order to help buiness users handle with organzational data, thus:

 - bringing inprecise, manually entered data, into a structrued format
 - looking up other attributes like hierarchy level, telephone number etc.

The more names I have to match in this way, the more powerful this feature is of course. On the other side, if the user has to click through hundreds of cells and choose the right outcome manually, the value is significantly lower.