1 of 1 people found this helpful
The language that Tableau uses is called VizQL, which tableau uses to transform the data into visuals. It follows some of the principles that Pat Hanrahan (founder of Tableau) used at Renderman. If you want to read more about the engine, you can refer to this paper: https://research.tableau.com/sites/default/files/sigmod2015-pawel-tableau9.0.pdf and this Core Tableau Technology: Data visualization and analysis
In terms of LODs, I believe your answer was ok. LODs are used to do calculations between different level of detail (granularity). However, I think it is important to explain the different LOD calculations too when answering a question like this during an interview. If you want to read more about it, you can see it here Overview: Level of Detail Expressions - Tableau
Hope this helps, if does please mark it as helpful so other users can refer to it.
Thanks for the reply by pointing to resource on Tableau Engine.
Went through it and was helpful to understand the concept and history about it.
Regarding the LOD question, yes I did mentioned the 3 types of LOD's, however I was not able to answer the question as he was looking why do we use LOD if we can perform the same analysis using complex calculations and also we cannot use the LOD expressions in Cube data source.
Any particular reason to go with the LOD expressions rather than calculations?
You are right! There are a few other things to consider about LODs.
1) Order of operations: Tableau's Order of Operations - Tableau
Fixed LODs are calculated among the frst things, therefore they can be used as opposed to context filters and that results in an increased performance.
2) LODs are always gonna give the expected result. Most of the times people would prefer to use Table Calculations instead of LODs but Table Calcs work with the things that are displayed in your view, whilst LODs refer to the data. Therefore, if you have a table calculation and someone changes something in the view, the calculation will change with it. In contrast, with a FIXED LOD calculation will remain the same.
Probably there are some other things but can't think of it right now. Definitely an interesting topic that is many times overlooked. Try searching in google "LOD The Data School" there may be some posts about it.
Hope this helps,
1 of 1 people found this helpful
Hard to tell what is in the mind of an interviewer - the explanations in the Help only go so far -
Think of an LOD as creating an additional layer (virtual) in the data - the layer works at a higher level than the granularity of the raw data (kind of like a hierarchy)
The higher level is not an aggregate so it can be used in all other calculations and in filters - sets and the like-
when you bring the LOD to the viz the underlying table for the sheet is now at that level - see below
this is just a simple LOD fixed at the cat level and the underlying table for the viz - the values in the table are at the higher level of the LOD
the lod can be used in table calcs (or other calcs) but you are now higher up the hierarchy
just a simple running total
and this is what the underlying table looks like
does that help
If this posts assists in resolving the question, please mark it helpful or as the 'correct answer' if it resolves the question. This will help other users find the same answer/resolution. Thank you.
Nice to connect with you again.
Thanks for the detailed explanation with the demonstration.
Agree with you it's really hard to give the exact answer the interviewer is looking for.
But the question remains we can achieve this using calculation as well right.
Keeping the post as unanswered, so that if anyone else wants to contribute to the post.
Sure - sometimes interviewers are just trying to see how deep the understanding goes - and not looking for a specific answer - particularly true when they pose open ended questions