SAS

[sasguru]

Not true at all. There are in fact lots of competitors, some of which are equal in capability, lower in licensing cost, better in terms of maintainability AND are cheaper to find skills for in the marketplace. There isn't much of a shortage of SAS developers (although there IS a shortage of ones who are also capable of having sensible conversations with the business without offending them with their arrogance!).

The other problem with SAS people is that they tend to suffer from Hammer-Nail syndrome.

What alternative do you suggest for clinical trials in pharma, for example?
Or creating models for underwriting in the insurance industry?
Or risk analysis in banking?
I don't do SAS development, which is in any case low level manipulation of data, prepatory to doing more interesting things with it.
If you want to do linear or logistic regression to takes some simple examples, SAS is virtually the only tool used commercially.

But like I said you low level techies haven't a clue about high level BI.

[OwlHoot]

Most posts on this thread only managed to reveal an unsurpassed depth of ignorance about what SAS ...

Must admit I may have been confusing it with SAP.

Anyway, it's the workflow system I was talking about, what ever that is called.

[Freamon]

What alternative do you suggest for clinical trials in pharma, for example?
Or creating models for underwriting in the insurance industry?
Or risk analysis in banking?
I don't do SAS development, which is in any case low level manipulation of data, prepatory to doing more interesting things with it.
If you want to do linear or logistic regression to takes some simple examples, SAS is virtually the only tool used commercially.

But like I said you low level techies haven't a clue about high level BI.

R? SPSS? Mathematica?

In fact, commercially, a lot of this stuff is done in excel (sadly).

[sasguru]

R? SPSS? Mathematica?

In fact, commercially, a lot of this stuff is done in excel (sadly).

Like I thought you don't have a clue

Do you think you can use SPSS and Mathematica, let alone Excel, with hundreds of millions of records?

To a certain extent I've used all the tools you've mentioned, horses for courses and all that,
but SAS scales up better than any others.

Now I'm going to start working with billions of records, I'm looking at Hadoop.

[AtW]

Now I'm going to start working with billions of records, I'm looking at Hadoop.

I am just writing SKA code that will deal with a trillion records.

HTH

[Freamon]

Like I thought you don't have a clue

Do you think you can use SPSS and Mathematica, let alone Excel, with hundreds of millions of records?

Ah, so if high volumes aren't an issue (which in most cases they aren't) then other tools are viable? How about a solution where you use a cheap commodity tool to cut the volumes of data down to filter out the few million records that are actually needed, then using some SAS alternative to process them?

To a certain extent I've used all the tools you've mentioned, horses for courses and all that,
but SAS scales up better than any others.

I thought SAS was the only tool used commercially? Obviously it turns out that isn't the case. SAS can handle higher volumes than some tools, but SAS itself has scalability limits. Telecoms, for example, is an industry where you're often dealing with data volumes that SAS won't handle.

You'll find a lot of environments where SAS is in use are a bit of a "closed shop" as they have so many SAS focused people that they don't bother to look at other (often very viable and far more sensible) alternatives. They get SAS consultants in, who obviously aren't vendor agnostic, and then just blindly accept their view that SAS is the "only solution". This is what I meant by Hammer-Nail syndrome.

Now I'm going to start working with billions of records, I'm looking at Hadoop.

Google are actually starting to phase out Hadoop as it is being superseded by other things...

[sasguru]

Ah, so if high volumes aren't an issue (which in most cases they aren't) then other tools are viable? How about a solution where you use a cheap commodity tool to cut the volumes of data down to filter out the few million records that are actually needed, then using some SAS alternative to process them?


I thought SAS was the only tool used commercially? Obviously it turns out that isn't the case. SAS can handle higher volumes than some tools, but SAS itself has scalability limits. Telecoms, for example, is an industry where you're often dealing with data volumes that SAS won't handle.

You'll find a lot of environments where SAS is in use are a bit of a "closed shop" as they have so many SAS focused people that they don't bother to look at other (often very viable and far more sensible) alternatives. They get SAS consultants in, who obviously aren't vendor agnostic, and then just blindly accept their view that SAS is the "only solution". This is what I meant by Hammer-Nail syndrome.


Google are actually starting to phase out Hadoop as it is being superseded by other things...

You're just blowing hot air and showing your ignorance now. You clearly don't understand why companies would want and need to process hundreds of millions of records.
Telecoms don't use SAS? Funny that I made loads of dosh contracting with telecoms in the past.
You haven't mentioned any alternatives to SAS for statistical processing and mathematical modelling on millions of records - although you have mentioned Excel
That just says what sorts of titchy companies you contract for.
Thanks for playing with the big boys, now go back to your javascript coding, there's a good fellow

[AtW]

Funny that I made loads of dosh contracting with telecoms in the past.

Was it in the 90s well before you've become a permie and out of boredom joined CUK?

[MarillionFan]

I am just writing SKA code that will deal with a trillion records.

HTH

Select count(*) from tblLinksNotFoundBySKA

[realityhack]

Must admit I may have been confusing it with SAP.

Anyway, it's the workflow system I was talking about, what ever that is called.

Yep - same here until I thought twice, then compared it to enterprise wofkflow, then googled.

Fair play - I'm a designer not a mathematician, each to their own.