Visual level filter changes reported data in line and clustered column chart

Ah sorry, no can do because it's cube-data, MSOLAP.5 . I forgot to mention that but now that I see your example showing the correct numbers, I realize that might be important.

After seeing your table I reckoned it might be helpful to show my own as well, and found another strange event occurring: The grand total does not decrease with 15, but only with 11 when enabling the filter. This got me thinking, maybe there is something going on with the unique (non-cumulated) values on which powerbi bases the total. A little background on the dataset: it's a count of preregistrations for an event, on which the bars in my previous post show the registrations per day, and the line shows the cumulated count of all registrations to date.

Although the CumulatiefAantalGeregistreerdOpleiding is calculated  (and I realize this might be related), i.m.o. a 'visual filter' should never alter data, just show different parts of it to emphasize on what is important. When I don't use that filter, the trail on the left becomes very long as there is a long period where almost nothing happens, as first registrations are most of the time about 5/6 months before the event. This automatically makes the important bits too small. I am currently using an excel template (which is a bit more versatile and lenient in visualisation ways) to acquire the data and distribute the results via email, but I want to take advantage of our company's enterprise powerbi license and give everybody continous access to the reports (and indeed, save some time for myself as well).