Hi, all here,
I am just wondering about the Discretization function available on analysis service server (which can not actually discretize data into user-defined expressions) . Isnt it redundant ? I mean since users can discretize their data more meaningfully based on their own expressions in database engine of SQL Server 2005.
Thanks a lot.
It is helpful in many circumstances. If you want a column discretized into statistically meaningful buckets, but there's no particular business reason for the breakdown, it is very useful. If there are particular business reasons for where the bucket divisions should occur, then you are correct, you should manually discretize apriori.
Eliminating a necessary preprocessing step for algorithms that don't support continuous values makes data mining easier for a wider range of people.
No comments:
Post a Comment