So I was thinking about this a lot of the holidays and have some thoughts.
1. One of the biggest reasons I like to start these discussions (Even if it comes at the expense of me sounding like an idiot) is that I get to see things from other people's perspective. It allows for discussion/debate that helps all of us learn, because not everyone's view or situation is the same. After reading all of these great responses, I can now see things from a perspective I didn't before. And I believe my opinions on the matter have changed for the better.
2. I think one of things that could be possibly holding this up is how it will affect Tableau's query process. For Extracts, I see less trouble because (theoretically) updating parameters could happen AFTER the initial Extract Query is fired off. as a part of a POST Query process. But I think there are issues with using LIVE connections, as Tableau might need to send back Multiple Queries to the DB (Both for Desktop and Server) prior to the Initial Query(ies) issued for the Viz (or Dashboard), since Calcs, would need the Parameters to be materialized first (This may need to happen every time a viz is loaded, since using a Live Connection could change what values are loaded into a Parameter). This could cause performance issues depending on the Grain of the Parameters. And we also have to consider Parameters that are based on Calculated Fields (This doesn't seem like much, but it could add a level of complexity).
3. I like what Jonathan Drummey mentioned regarding Mental Models, and how Tableau is different than other reporting solutions. Though I would love for Tableau to release documentation on it, I'm wondering if this could be a good topic for a Think Data Thursday (Patrick Van Der Hyde )?
4. I see a lot of people pointing to the fact that the term Dynamic Parameter is largely "undefined" (Or rather interpreted in many different ways). It might be good to have an open discussion between Tableau and the Community on Dynamic Parameters so that we can address/define what Dynamic Parameters actually mean, and perhaps create formal documentation based on those discussions.
If I have said anything incorrect, or miss-leading, please, PLEASE, let me know. I would much rather learn WHY I am wrong and learn, than continue having thoughts that may not accurate.
Best regards, and thank you for everyone who has participated thus far!
Dynamic (autoupdating) and multiselect parameters for me would allow me to do comparison charts on bigdata (which rules out datablending). I'm working for big corporation and they are very surprised that tool of their choice is not capable of doing what they want (to select multiple items, select another multiple items in the same dimension - not necessarly overlaping first set and compare their measures). To be honest, I share same feeling. This is quite basic and pretty common use case which is impossible to do currently without datablending workaround (which is not viable for bigdata).
Maybe highlighter showcased on TC15 would solve some problems but definitely not all in this use case.
For comparison charts it would also be very useful ability to combine two charts (based on two different datasources and their own set of filters) together in single one. Similar to Dual Axis functionality but again based on two separate datasources and separate set of filters.
We have put significant amount of work on performance side of our reports. Basically we have come to a point where filters are our last bottleneck but amount of them and way they work are set as mandatory requirement for business and we cannot change it.
We found that setting all joins in tableau and checking Assume Referential integrity speeds up our reports significantly. Fillup filter queries were very fast. But overall report performance was still not comfortable to the end users. In bigdata world joins are very expensive so we went for denormalized tables. Now main query is fast but filter fillup queries are again killing performance.
Currently as a workaround to performance problems with many "Only Relevant values" filters, lack of hierarchical filters , we are building GUI with filtering part of reports and use tableau only to display results in embedded worksheet.
You can imagine how much more effort and design time is needed to achieve what could be easily achieved by dynamic (autoupdating on separate datasource) and multiselect parameters.
Not to mention we are hearing more and more often a dreadful question: why we are still using a tool which limits us so much?
My reply "Hold on little longer. It is most desired functionality and Tableau is working on implementing it" does not work anymore.
1 of 1 people found this helpful
Great post Rody. Many of the multi-select comparison needs that you describe are now solved with Set Actions. Find out more here: 8 ways to bring powerful new comparisons to viz audiences with Tableau set actions | Tableau Software