what is the best way to determine numeric patterns within a dataset

[D]

I have normalized data in sql server in a table like (testset, testnumber,
result) . There can be 5-10 tests in a testset and anywhere from 1500 rows
to 100000 rows of testsets. I need to start searching for patterns within
the data such as the total number of times 15,16 & 17 occurred. My first
thoughts were to write a stored procedure but that hasn't proved to be the
best as I am still trying to get that to work.

In the essence of speed would it be better to do this analysis on the c#
side or continue within my efforts of using stored procedures? I have even
considered a third party library as more indepth analysis may be required.

Your thoughts on the matter are truly appreciated.

Regards
Dave

[Salvador]

Hi,

Use a Stored procedure to make the calculations, if you have to do in c# you
have to load data in memory or using a datareader, at the end you r adding
more unnecessary layers.

regards
Salva
"D" wrote:

I have normalized data in sql server in a table like (testset, testnumber,
result) . There can be 5-10 tests in a testset and anywhere from 1500 rows
to 100000 rows of testsets. I need to start searching for patterns within
the data such as the total number of times 15,16 & 17 occurred. My first
thoughts were to write a stored procedure but that hasn't proved to be the
best as I am still trying to get that to work.

In the essence of speed would it be better to do this analysis on the c#
side or continue within my efforts of using stored procedures? I have even
considered a third party library as more indepth analysis may be required.

Your thoughts on the matter are truly appreciated.

Regards
Dave

[D]

ok, I had initially thought that too but with the pattern searching I've
been trying to accomplish in t-sql I was starting to think that it might be
done faster in somewhere else. Speed is of the essence and some of these
queries don't seem that fast.
"Salvador" <Sa******@discussions.microsoft.com> wrote in message
news:1C**********************************@microsof t.com...

Hi,

Use a Stored procedure to make the calculations, if you have to do in c#
you
have to load data in memory or using a datareader, at the end you r adding
more unnecessary layers.

regards
Salva
"D" wrote:

I have normalized data in sql server in a table like (testset,
testnumber,
result) . There can be 5-10 tests in a testset and anywhere from 1500
rows
to 100000 rows of testsets. I need to start searching for patterns within
the data such as the total number of times 15,16 & 17 occurred. My first
thoughts were to write a stored procedure but that hasn't proved to be
the
best as I am still trying to get that to work.

In the essence of speed would it be better to do this analysis on the c#
side or continue within my efforts of using stored procedures? I have
even
considered a third party library as more indepth analysis may be
required.

Your thoughts on the matter are truly appreciated.

Regards
Dave