PASW: Where is SPSS going?

[Richard Ristow]

So it's now "predictive analytic software"

 From SPSS's promotional descriptions, 'predictive analytics' sounds
like some is specialized predictive techniques; from the
descriptions, I can't see what these are.

Does anybody on the list do what SPSS, Inc., refers to as predictive
analytic projects?  What problems are encompassed under this
description, and what techniques? If you do predictive analytic, what
are the steps in such a project - needs? data? models? presentation?

Now, I worry, and perhaps many of us worry, about this change in
SPSS's self-description. For these thirty years or so, I've used
SPSS, and taught it, as powerful and flexible, for data preparation
and analysis for a wide range of statistical problems. That's what
the membership and traffic on this list feels like.

I'm sympathetic with SPSS, Inc.  It's much easier to market a product
if you say, 'It does this thing, and this is what that thing will do
for you', than to say, 'this is a very flexible product for a whole
lot of things.' Some of the analysts, and anybody consulting on
statistical problems, wants the latter; but most managers don't.

But, if SPSS, Inc., is emphasizing one class of its applications, its
other features will eventually be neglected. That's just about a law.

So, should the rest of us, most of us on the list, make definite
plans to migrate from SPSS? How quickly? Would it be easier for SPSS,
Inc., if we did?

-With regards and concern,
  Richard

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[ViAnn Beadle]

Predictive Analytics typically provides techniques which provide predictions
about people's behavior. The techniques are suited to massive datasets and
relatively messy data (think data mining). Among them are classification
trees or logistic regression used to build scoring models for things like
propensity to buy, propensity to default on a loan, propensity to commit tax
fraud; and association models for market basket analysis which determine
offers for cross-selling, book suggestions in Amazon, Netflix
recommendations, etc.

An important component of any software suite in this area is a scoring
engine and all the glue required to provide a real-time score. A typical
fully integrated solution will cost more than a $M.

Plug "predictive analytics" into a Google search. You'll get about 800,000
hits. Also look at the kdnuggets web site for more information.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Richard Ristow
Sent: Tuesday, August 04, 2009 10:21 AM
To: [hidden email]
Subject: PASW: Where is SPSS going?

So it's now "predictive analytic software"

 From SPSS's promotional descriptions, 'predictive analytics' sounds
like some is specialized predictive techniques; from the
descriptions, I can't see what these are.

Does anybody on the list do what SPSS, Inc., refers to as predictive
analytic projects?  What problems are encompassed under this
description, and what techniques? If you do predictive analytic, what
are the steps in such a project - needs? data? models? presentation?

Now, I worry, and perhaps many of us worry, about this change in
SPSS's self-description. For these thirty years or so, I've used
SPSS, and taught it, as powerful and flexible, for data preparation
and analysis for a wide range of statistical problems. That's what
the membership and traffic on this list feels like.

I'm sympathetic with SPSS, Inc.  It's much easier to market a product
if you say, 'It does this thing, and this is what that thing will do
for you', than to say, 'this is a very flexible product for a whole
lot of things.' Some of the analysts, and anybody consulting on
statistical problems, wants the latter; but most managers don't.

But, if SPSS, Inc., is emphasizing one class of its applications, its
other features will eventually be neglected. That's just about a law.

So, should the rest of us, most of us on the list, make definite
plans to migrate from SPSS? How quickly? Would it be easier for SPSS,
Inc., if we did?

-With regards and concern,
  Richard

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Hector Maletta]

I am not sure that doing something specific is more marketable. Imagine MS Word being marketed as a software for preparing business letters, or philosophical essays, or whatever, and not as a general-purpose word processor. Imagine MS Excel being marketed as a tool for preparing budgets or solving linear programming problems. Or Google as a website to find suitable hotels.
SPSS is a general-purpose statistical software, and that´s what it is, no matter what fancy name they put on it.
Hector

----- Mensaje original -----
De: Richard Ristow <[hidden email]>
Fecha: Martes, Agosto 4, 2009 11:21 am
Asunto: PASW: Where is SPSS going?

> So it's now "predictive analytic software"
>
> From SPSS's promotional descriptions, 'predictive analytics' sounds
> like some is specialized predictive techniques; from the
> descriptions, I can't see what these are.
>
> Does anybody on the list do what SPSS, Inc., refers to as predictive
> analytic projects?  What problems are encompassed under this
> description, and what techniques? If you do predictive analytic, what
> are the steps in such a project - needs? data? models? presentation?
>
> Now, I worry, and perhaps many of us worry, about this change in
> SPSS's self-description. For these thirty years or so, I've used
> SPSS, and taught it, as powerful and flexible, for data preparation
> and analysis for a wide range of statistical problems. That's what
> the membership and traffic on this list feels like.
>
> I'm sympathetic with SPSS, Inc.  It's much easier to market a product
> if you say, 'It does this thing, and this is what that thing will do
> for you', than to say, 'this is a very flexible product for a whole
> lot of things.' Some of the analysts, and anybody consulting on
> statistical problems, wants the latter; but most managers don't.
>
> But, if SPSS, Inc., is emphasizing one class of its applications, its
> other features will eventually be neglected. That's just about a law.
>
> So, should the rest of us, most of us on the list, make definite
> plans to migrate from SPSS? How quickly? Would it be easier for SPSS,
> Inc., if we did?
>
> -With regards and concern,
>  Richard
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
> [hidden email] (not to SPSSX-L), with no body text
> except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Barnett, Adrian (DECD)]

Hi Richard
The best description I've seen of what IBM is trying to achieve by buying SPSS, and how this notion that what the software about is this rather silly-sounding (well, in the context of the interests of most of the people on this list) notion of 'predictive analytics' is this article:

http://www.theregister.co.uk/2009/07/28/ibm_bao_boxes/

It seems it fits into a bigger context - IBM is going after what used to be called 'the glasshouse', and SPSS fills a hole not previously covered.

It's been increasingly apparent that SPSS are no longer very interested in their traditional audience of people with beards and jeans because the big dollars are with the suits running the IT departments of big companies who till now have, I think, mostly been buying SAS when they need to do all that 'number stuff'. Their purchase of Cognos got them something quite nice for creating and easily distributing reports, but there was still a hole where it came to more sophisticated analysis. The suits seem finally to be realising that making intelligent use of their company's data requires more than buying a data warehouse and creating yet another sort of report.

It may well be, as you say, that with IBM in charge, the focus will move so thoroughly to this data mining direction that they really will no longer care about the more prosaic needs of the likes of this group. If their strategy works and considerably more income derives from these big IT shops, it's difficult to see how they would be able to cost-justify any developments that would improve things for the traditional SPSS audience if it did nothing for their new pals in the suits with the big cheque books.

I think our best hope is that the interests of the 2 groups are too closely bound up for one to suffer. I have no idea how realistic that hope is. (Personally I'm hoping IBM will bring some better quality control to the software development process.) I guess there's always R and Stata if things get too bad. We will just have to wait and see what happens.


Regards,



Adrian Barnett
Information Analyst
Educational Measurement and Analysis
Data Management
DECS
ph 82261080

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Richard Ristow
Sent: Wednesday, 5 August 2009 1:51 AM
To: [hidden email]
Subject: PASW: Where is SPSS going?

So it's now "predictive analytic software"

 From SPSS's promotional descriptions, 'predictive analytics' sounds
like some is specialized predictive techniques; from the
descriptions, I can't see what these are.

Does anybody on the list do what SPSS, Inc., refers to as predictive
analytic projects?  What problems are encompassed under this
description, and what techniques? If you do predictive analytic, what
are the steps in such a project - needs? data? models? presentation?

Now, I worry, and perhaps many of us worry, about this change in
SPSS's self-description. For these thirty years or so, I've used
SPSS, and taught it, as powerful and flexible, for data preparation
and analysis for a wide range of statistical problems. That's what
the membership and traffic on this list feels like.

I'm sympathetic with SPSS, Inc.  It's much easier to market a product
if you say, 'It does this thing, and this is what that thing will do
for you', than to say, 'this is a very flexible product for a whole
lot of things.' Some of the analysts, and anybody consulting on
statistical problems, wants the latter; but most managers don't.

But, if SPSS, Inc., is emphasizing one class of its applications, its
other features will eventually be neglected. That's just about a law.

So, should the rest of us, most of us on the list, make definite
plans to migrate from SPSS? How quickly? Would it be easier for SPSS,
Inc., if we did?

-With regards and concern,
  Richard

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD