visual filter -- unpredictable behavior

[RaidasN]

Hi everyone,

I found an issue with visual filters and was wondering if anyone came across with a similar problem.

I created a simple chart containing store name and conversion (volume divided by footfall). As we know in real life we can't get all the data we need so some of the footfall is 0. To avoid infinity (dividing by 0) results I applied a visual filter on footfall column to exclude 0 values. In theory, it should do the trick, but the answer I got was unexpected, to say the least. Not only it wasn't behaving the way I intended to, it created an additional error. Now my table had not only infinity values but NaN value as well. 

Eventually, I was able to get the answer  I needed, by adding footfall column into "Tooltips" and applying the same filter to footfall column as before (footfall is not 0). My chart seems to have correct data now, but I'm still confused how I can get different numbers by applying the same filter... 

P.S. the new filter works even if I remove footfall from "Tooltips". 

If anyone can give me an explanation I would be grateful. 

Regards,

Raidas N

[v-haibl-msft]

@RaidasN

I’m not very clear about your issue here. I try a simple similar scenario as you but everything seems to work well as below. Is there any difference between our scenarios?

If possible, you can upload your PBIX file to online file service like OneDrive and share it to me.

Best Regards,

Herbert

[RaidasN]

Hi Herbert,

Thanks for quick reply. Sorry, but I can't share the file because it contains confidential clients information. This post was mostly aimed to make Microsoft aware of the issue, not to find a solution for it. My method works so I am fine to leave it as it is for now.

I will try to describe differences between our scenarios in more detail.

Footfall, store data, and sales are stored in different tables and different granularities. For example, we are getting footfall by week, by store, but we are getting volume by day, by store, by product. So the first step is to create measures to aggregate everything to desirable granularity (week-to-date, month-to-date, etc... by store). Only then we can calculate conversion rate using our new measures. I already checked that all links between tables are active and correct. 

Regards,

Raidas

[v-haibl-msft]

@RaidasN

You can try to use https://www.mockaroo.com/ to create dummy values for the tables. If you can repro the same issue with these dummy values, then you can share it with no sensitive data.

Best Regards,

Herbert