I'm a newbie to OLAP and Data Mining so I might be leading you astray, but
it's similar to things I do outside of OLAP.
If I understand your problem correctly then this isn't an OLAP problem
directly but
rather a statistical or perhaps data mining problem. If you're not using
data mining software then you need to determine the normal level -- which is
often called the 'expected value' -- and then look for things that are
farther away from the expected value than the standard deviation would
predict. This is always formulated as a percent chance. e.g. on "Tuesday
there was a low value-- the percent chance that this low value is the result
of random fluctuations is 1%".
A statistics book can give you the formulas that you need but it would
probably be a good idea to investigate data mining before taking the trouble
of doing it yourself.
Jamie Beerbower
"Robin" wrote:
Quote:
How can you analyse information based on day or time of day to find any
exceptions to the normal level? |