Hi,

I'm wondering if there's any doc that describes the process flow of Build
(build |
deploy), Database (Process), Mining Model (Process) for the BI Development
Studio.

I built a decisioin tree model and attempted to process it but failed.
Here's a screen that bugged me, password prompt for impersonation account of
the data source in question (the database is hosted by the sql db
engine/service
NOT analysis engine/service), I entered pwd for 'sa' for it, it generated
err, msg to the effect, syntax for the ImpersonationInfo object incorrect.
A couple of questions here:
1) Is IIS required for BI Dev Studio and Analysis Service to work together?
2) Any pointer to white paper or article on the whole process of creating
mining model and use it
with sql server 2005? The tutorial seems inadequate to me.

Thank you very much.

Never mind. I solved the problem. Talking to myself: when doc is inadequate
or the like, use logic, use logic, repeat after me, everyone

"ImJoe" wrote:

The impersonation information determines what user Analysis Server
impersonates on the thread it uses to establish a connection to the data
source (using the connection string specified in the data source, of
course). If you're not using Windows (integrated) security and your
connection string contains a SQL username (like "sa") and password, the
impersonation information is not essential - but it's still important from a
security perspective to set it to a safe value since it still determines the
security context the connection is running under (and you may change the
connection info later to switch to integrated security).

IIS is not required for BI Dev Studio to connect to AS - the connections
are made through one of the object models (depending on the context) - AMO,
ADOMD.NET or OLE DB - all of which internally connect using XMLA directly,
since that is the native protocol for AS2005.

--
-Raman Iyer
SQL Server Data Mining
[This posting is provided "AS IS" with no warranties, and confers no
rights.]
"ImJoe" <ImJoe (AT) discussions (DOT) microsoft.com> wrote

Thanks for the followup, RI.

Another question, it's on Lift.
The List Chart seems to require at least two models. imho, even if I have
only one,
and if the List Chart function can support that it would be beneficial to
see the Lift degree

IDEAL
model
GUESS

Another related question, for INPUT TABLE, how can I modify a model to allow
selection of more tables instead of one?

TIA.

"Raman Iyer [MS]" wrote:

You can select one or more models in the lift chart. I think you might be
confused by the two types of lift charts - one where you select a target
value and the one where you don't. When you select a target you get a
cumulative gains chart which is very familiar to data miners. You can get
whatever statistics you want by clicking on the chart and looking at the
mining legend.

A model has more than one table if it has nested tables. Download the movie
click sample data and project from www.sqlserverdatamining.com for an
example.

--

-Jamie MacLennan
SQL Server Data Mining
This posting is provided "AS IS" with no warranties, and confers no rights.
"ImJoe" <ImJoe (AT) discussions (DOT) microsoft.com> wrote