Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I use Tableau almost daily. Initially I found it harder to work with and avoided it for a while. On 2nd attempt, for some reason, everything clicked into place and I now find it fairly intuitive. Some of my opinions/experiences about it:

- Beyond simple queries, it becomes more easier to setup a view or two behind Tableau than to do it purely in Tableau. E.g. multiple level of latest times, more complex aggregations etc are more easily done in SQL than via Tableau.

- For some things Tableau doesn't do what we typically expect. Couple of examples:

- If you run a custom query joining two tables with some of the column names being same in both, Tableau cannot deal with it. Setting alias for columns using "as" works fine in SQL so I think it is not unreasonable to expect that to work.

- I recently tried to chart lag in request/response in microseconds. The data is stored as TIMESTAMP in our Vertica database. However, I found that Tableau doesn't show time more granular than seconds. That is kind of odd in today's common low-latency high-performance technology environment

Having said that, I really like the features it offers viz. Calculated fields which don't go away just because I change the data source, drag-and-drop dashboard constructions, very nice visualizations etc.



What do you use it for?

If it's for analysis why not use python or R, way more powerful and full of tools & libraries that give you speed you need?

If it's for reporting, why not to use chartio, looker or anything that connects directly to your db?

My opinion: If you know some python or R, you might still need chartio, looker and similar but Tableau doesn't make sense anymore.


I wasn't part of the decision making effort but I believe Qikview was also considered but the analysts preferred Tableau. It is used by the analysts as well our for our internal monitoring/charting/analysis purposes.


I see. I've also found myself in a situation where the company was using it and I had them drop it.

For reporting the idea of transforming & moving the data around really doesn't make much sense when compared to solutions which directly leverage your db or at least create an unique consistent datasource instead of many so-called extracts scattered around.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: