Subscribe to DSC Newsletter

What does IBM’s acquisition of SPSS mean for the analytics field?

On Tuesday, IBM announced that it was acquiring SPSS for 1.2B in an all cash deal. Not surprising that one of the big database companies would go after a data mining software company. For many companies, it’s all about the database. IBM’s existing Intelligent Miner package is not widely used, though I’ve heard that it’s powerful. (Never used it myself.)

My question is, what does this mean for the analytics field / to the analytics community?

My guesses:

Microsoft will continue to chart their own course. Not a wise move in my opinion. SQL Server Analysis Services is IMHO, fairly awful and has very few features. Plus, Microsoft has widely known (and academically published) problems with creating accurate statistics routines. (e.g. McCullogh & Wilson, On the accuracy of statistical procedures in Microsoft Excel 2000 and Excel XP, Computational Statistics & Data Analysis 40 (2002) 713 – 721)

Oracle bought Thinking Machines International and their Darwin data mining suite in 1999, and Oracle Data Mining is somewhat widely used and fairly powerful. They might continue to chart their own course as well.

But if Microsoft or Oracle did go after someone, I’m guessing StatSoft would be the target…

Thoughts?

Views: 1065

Reply to This

Replies to This Discussion

I am not optimistic about the future of SAS. Their database technology is obsolete, far behind Oracle or SQL-Server, their statistics cannot even compare to R, and you cannot easily interface with R from SAS EM, as you can for example from STATISTICA Data Miner, although Statsoft's Data Miner has more statistics than SAS EM.

Do they hope that R will go away? SAS' strength is a bunch of domain-specific macros that they sell for arm and leg as specialized applications and force us to buy them again and again every year!!! These days the 65% annual fee is not something users will put up with much longer. I am not sure how long SAS will get away with their current business model in this competitive environment. BTW someone here has his math wrong, if SAS was purchased for 4x sales, the price would be 8 billion, not 80 billion; 80 billion is a GNP of a good size country.
Well, their database technology is not their flagship product and I don't think many will dispute that DB2, SQL Server etc are dominant players in that arena.......however they have a formal partnership with Teradata that gives them leverage and is available to people that want to take it to the next level.
Then look at other "statistical" companies like SPSS for example. Their "database" capability is far inferior to SAS and in some other companies' cases non-existent. In case of SPSS that will I suppose change now that they are owned by IBM.
You can use R thorough SAS/IML I believe and SAS EM will allow you to import whatever custom models you want.
Again you're right that R is second to none in the statistics arena but R is not a commercial software. There are no implied warranties and guarantees that some anonymous dude across the globe wrote an R program does exactly what it's supposed to do or it says it does.
It is "as is" kind of a deal.
Big corporations will not go for that no matter how "cool" the algorithm is with some exceptions and specific cultures like say Google.
They've been other perhaps more stable and better open source operating systems, but Windows is still here to stay and dominates corporate culture.
I don't see bank or pharma industries abandoning SAS any time soon.
Statistica looks promising but its interface, look and feel need work. It does promise to be
a worthy future competitor in the market.
You've forgotten other SAS offerings like SAS/ETS for forecasting including the server version, SAS Enterprise Guide that is far more complete and useful than say Business Objects or Cognos at the moment, SAS/OR for Operations Research, SAS Fraud software, SAS JMP for advanced visualization, SAS/GIS and others. These are much more than "macros".
I can't think of any other company that is competitive at that level in pretty much every aspect of analytics, from BI to database, statistics, data mining to forecasting and newly launched fraud detection. Most of them are big players in one or two of these arenas (i.e. SQL server in database) but none have SAS versatility.
I agree that their license is ridiculous and price is too high. Perhaps with the emergence of new players they will revisit that as that's probably no. 1 reason why would one consider switching to other packages.
Yes, I agree, but because SAS database and database tools are somewhat "strange" and actually quite obsolete, they should do what some of their competitors do – open up to (connect directly to) high-performance data bases.
I use Statistica extensively because I need to hook up to various datasources/bases, and the IDP interfaces/tools that came with Statistica Data Miner are really nice because they allow you to read Oracle data directly without importing; it treats an Oracle or SQL database as if it was a native data file – very nice/fast! No comparison to accessing Oracle data from SAS, but SAS will never develop something like that, because they want to lock you into their own (costly) database infreastructure (trap?).
Yeah, I screwed up on annual sales: $2B (ref: D&B numbers from Manta). At any rate, I think my point remains that SAS would make for a difficult acquisition target. On the other hand, sometimes big bold moves like that can pay off – Oracle acquired Peoplesoft for $10.3B and it was probably the smartest thing Oracle ever did.

SAS is virtually institutionalized in pharma, holds the majority of the academic market (I’m guessing), and has large stakes in fraud and compliance. Those are SAS’s strengths. Their DB, errr. To say that it’s not their flagship product is putting it very nicely, but SAS tends to play well with other databases. I don’t dislike SAS, other than pricing/licensing, but I do like some other things better. Other than licensing, SAS’s other big Achilles heel is that it lacks flexibility.

I might argue with the characterization that R is second to none in the statistics area. I’m a huge fan of Stata, which gives me the flexibility of R without some of R’s big drawbacks. R isn’t bad software, but it’s got some faults and it’s not exactly oriented toward productivity… BTW, there are some commercial versions of R that have been validated (e.g. REvolution R).
I don't know what you mean by flexibility, but you can utilize R through both SAS and SPSS.
As well, you can also write SQL in SAS using PROC SQL procedure.
Yes, because SAS is just a programming language and you can of course program interfaces to other applications; I know that there are also import/export facilities to move data in and out of SAS.

Still, the SAS programming language is a hold-over from the mainframe days, it does not support objects!. All modern software uses object-oriented languages which make building and maintaining applications, and reusable parts much easier (by the way, the Statistica object model that I use is very well thought out and organized, and it based on standard VB syntax! It feel that SAS is behind and out-of-step with where modern software technology is headed, and my recommendation is to use standard data base systems and languages (SQL) to build a robust, fast data warehouse; then use the best tools out there to operate on the data, to analyze data, and so on.
SAS basically restricts you to the statistical routines that they’ve developed. If you license Proc IML, sometimes you can cobble something of your own together, but that’s the limit. So there’s no real way to go off on your own to follow, say, the latest time series models, or simulation models in inventory management, or…

Yeah, you could interface to R, but that’s kind of a “so what?” sort of deal. I could just use R to begin with. (I personally choose not to.)

I’m not trying to pick on SAS. I used it for many years - on mainframe, Unix, and PC. I appreciate that it has native drivers for the major databases instead of making you resort to ODBC. It produces really nice 3D graphics. It’s got some really nice features; I just choose to use something that gives me some freedom to chart my own course.
A side thought on Statistica - Statsoft isn’t SAS, the SAS business model doesn’t translate well to Statsoft, and they really need to put their pricing structure up on their website.

I mean, the reason a SAS salesperson has to give you the quote is so that someone is available to call an ambulance in case you have a heart attack after hearing the price…
BTW, how large is StatSoft? I cannot find any info on their web page?

They must be a large company - they have 35 offices on all continents and a Google map that gets you directly to every office, neat.
Manta says $20M/Year. I'm not 100% confident in Manta's numbers, but they're generally in the ballpark.

http://www.manta.com/company/mmg3l8s
No way! missing a zero? StatSoft is everywhere, they must be at least 200M/year.

RSS

On Data Science Central

© 2019   AnalyticBridge.com is a subsidiary and dedicated channel of Data Science Central LLC   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service