Skills for a Data Analyst

[Bagpuss]

back to the question... Doesn't it depend on how you define a data analyst?
If it's someone using the db to produce low level listings and reports then maybe yes. If it is someone using data in developing sophisticated models and then producing reports/presentations/technical documents then I would no more expect them to know about databases than how to unblock the company toilets, it's not their job, unless we are talking a very small company.

[MrRobin]

But to produce those sophisticated reports and presentations etc you need to be able to aquire the raw data to begin with... okay, some dba could send you a data dump every time you need something new but it would be far more efficient if the analyst could do this themselves, and even taking it a step further by integrating dynamic queries into their reports so they can be updated with a click of a button.

[BoredBloke]

A true contractor would have said they could do it in two days, done a script to do it in a few seconds, and spent the rest of the time on CUK.

Which is exactly what I did - Mine did it in seconds but I only streamlined* it just before I left and handed it on.

* told them that I found a way to significantly increase its performance.

[Bagpuss]

But to produce those sophisticated reports and presentations etc you need to be able to aquire the raw data to begin with... okay, some dba could send you a data dump every time you need something new but it would be far more efficient if the analyst could do this themselves, and even taking it a step further by integrating dynamic queries into their reports so they can be updated with a click of a button.

In the industry I work their are colleagues who's job it is to create datasets from raw data, this involves cleaning of the data, setting up queiries and transposing the data into useable formats. Having 'data management' is to prevent noddies bypassing local SOPs and extracting unclean and unvalidated data. Maybe in other industries the procedures are more relaxed, and less regulated.

[Sockpuppet]

I want to learn SQL server and the like but struggle to find the time. I use SQL in my VBA but I cheat as I tend to build the query in the query builder and then hack the resulting SQL statements so that I can throw variables at it in VBA.

Nowt wrong with that.

I do that for the simpler queries. Unfortunatley when you start to head towards nested sql and pivot its gets harder.

[Sockpuppet]

back to the question... Doesn't it depend on how you define a data analyst?

If it's someone using the db to produce low level listings and reports then maybe yes.

If it is someone using data in developing sophisticated models and then producing reports/presentations/technical documents then I would no more expect them to know about databases than how to unblock the company toilets, it's not their job, unless we are talking a very small company.

Hmmm...

I'd expect a decent data analyst to be able to interpret the data and understand what it means as well as demonstrate how they collected it.

I have to do pretty much most things from collect the data to analyse it to tell people what it means and how it affects X. Ok, that's why I am a contractor, I'm expected to do all that stuff. But even on project when I get given permie assistants I want them to know where it came from and then what it means.

Ok, I'm not expecting them to be able to go into a datawarehouse and know how to write a massive SQL list but if someone send them an access file I'd want to know they can write simple SQL to produce summaries of the data so they can change them quickly rather than having to re-import into excel and mess around with pivot tables.

For example a client I do work for has recorded the axle weights of vehicles leaving its gatehouse(s) and wants to do a study of actual weights. Road transport law is a pita so I'd not expect to have to take a day to explain to an analyst in a logistics company what his criteria and banding should be. He works for a logistics company so should know this stuff.

[SandyDown]

I would expect a data analyst to have an understanding of data modelling, and data structures. Also ability to connect and view the databases (perhaps through a GUI, or data view – most databases have that) so firing data queries can be done visually. As well as understanding data quality and troubleshooting data problems.

I don’t expect a data analyst to be sitting in front of a Unix machine firing some PL/SQL complex type of queries, they can always ask a techie to do that for ya as long as he/she design a query for them(give the criteria).

[thunderlizard]

I think SQL was originally invented so that data would be easily accessible to accountants, researchers, engineers etc. who wouldn't know Data Control Language. So a data analyst should be able to use it: or at least shouldn't complain if his job required it.

[expat]

I think SQL was originally invented so that data would be easily accessible to accountants, researchers, engineers etc. who wouldn't know Data Control Language. So a data analyst should be able to use it: or at least shouldn't complain if his job required it.

COBOL, FORTRAN, RPG, SQL ..... all invented to eliminate the need for programmers. All have given me a programming income at one time or another. Bring on another attempt to eliminate programmers, I say.

[Bumfluff]

Agreed they should be handy with SQL, however if they are female with a nice body I could forgive the lack of SQL