Forum Discussion
Prove this wrong: this simple Tableau visual cannot be made in Power BI?
I'm not trying to be argumentative; I am certified in both so I'm not anti-Power BI. But I am first a SQL developer, then learned Tableau, and then Power BI. If you SAY that this can be made in Power BI, I need you to make it. Because I have tried to use the CALCULATE statements, etc. Even google AI will give me directions to make this visual in Power BI, but it's wrong and doesn't actually produce what I am looking for. This is really a very simple visual (bar graph of a sub-dimension with an average line at the dimension level). This is the superstore.csv dataset. It's very simple and should be searchable and downloadable anywhere.
Hopefully this will intrigue because it sounds like a challenge, but of course if I am wrong I am happy to be wrong if I learn how to do this. In fact, I'm kinda surprised if I am indeed correct... this visual is simple and applicable in real life. Using DAX is of course ok/welcomed. I can envision a possible way by adding an entire second copy of the data, turning it into a miniature table in the data model to pre-calculate the average at the region level (?), but that seems like a stretch and I'm hesitant to be "ok" with that.
Hi cmd1218
It looks like you didn’t actually read what I wrote.
The points I raised are not about statistics but about effective data visualization and UX, which rely heavily on cognitive psychology.Your remark about “why would a user want to see the X-axis” is odd, because I explicitly mentioned that one of the problems with the Tableau visual was the unreadable X-axis labels (rotated vertically instead of horizontally).
At the end of the day, whether you trust my UX assessment is up to you. I can’t force you to “drink from the well,” but the technical fact remains: a Tableau-style visual like that does not exist natively in Power BI.
And if your real-world scenario is different from the toy example you posted, no problem—please share a more realistic example so the answers can be more relevant.
If this post helps, then please consider Accepting it as the solution to help the other members find it more quickly
9 Replies
No interest in proving right or wrong; only to say, it's simple enough to author such a measure.
Hi cmd1218
Technically, this type of visual doesn’t exist as a core visual in Power BI (unless you build it with Deneb or another custom visual).
The solution suggested earlier is problematic from a data-viz and UX standpoint:
-
Two Y-axes distort proportions and make comparisons unreliable.
-
Continuous line is misleading because the brain reads it as a time trend, which is not true here.
-
Small multiples are not suitable for this case: they show different hierarchies and also render empty categories, which adds clutter.
Also, the Tableau version is not really optimal if the intent is to:
-
Compare performance across regions.
-
Compare performance by state.
-
Understand each state’s deviation from the regional average.
Why?
-
Too many colors cause cognitive overload, nothing gets highlighted.
-
The average line is different for each region, so users can only compare within one region, not across regions.
-
Vertical axis labels make reading harder.
A much more effective approach in Power BI is to use a table with integrated bar/variance visuals, showing:
-
Sales per state.
-
Delta vs. the regional average.
-
Clear reference to the average by region.
This allows instant ranking, spotting deviations, and filtering by region when needed. Cleaner, easier, and more actionable than either of the previous visuals
The pbix with the example is attached
If this post helps, then please consider Accepting it as the solution to help the other members find it more quickly
I appreciate the time you put into your reply.
The Tableau version is entirely correct from the view point of a statistician. It follows data visualization best practices (you are correct that the regions shouldn't be colored differently because they are already labeled across the top; I did this by mistake because, in our real data, "regions" are not labeled across the top and panes are removed because some "regions" have a TON of "states" and we use a horizontal scroll). As far as your points about "why would someone want to see x?", they don't hold because you don't know the actual meaning of the fields I use in real life. You also don't know the end user. My question was was correct in the beginning. "bar graph of a subdimension with average lines at the dimension level of detail."
-
No, it's incorrect. Your calculation is correct in theory, but it cannot be brought into the visual correctly. The line is wrong. It would take me a long paragraph to explain in full (I will if you want me to! really! i'm just in a hurry at the moment) but for the sake of brevity, the line in tableau is showing a discrete value at the region level and then of course a continuous value at the state level. Maybe a simpler reason is this: this shows a slope connecting regions?
