You design a SQL Server 2008 Analysis Services (SSAS) solution. You have the following requirements for a single data mining model:
-Group all customers by two different age ranges.
-Group all customers by ten different age ranges.
You need to design the model to meet the requirements.
What should you include in the design?
A.
one column of the Long data type and the Discrete content type
B.
one column of the Long data type and the Continuous content type
C.
two columns, each of the Long data type and the Discrete content type
D.
two columns, each of the Long data type and the Discretized content type
Explanation:
Tip: Group by 2 ranges agrupments = "Discretized"We do this by choosing Discretized as the Content Type, and then selecting a method that groups continuous values into a discrete number of buckets (that is, ages 1115, ages 1620, and so on). Often, it is much easier to do analysis on discretized values than on continuous values. For the purposes of a given analysis, the buying habits of 1620-year-olds may be similar enough so that we can study them as a group in one discretized bucket.
(McGraw-Hill – Delivering Business Intelligence with Microsoft SQL Server 2008 (2009))Discretized The column has continuous values that are grouped into buckets. Each bucket is considered to have a specific order and to contain discrete values. You saw an example of this in Figure 12-2 using the Age column in the Targeted Mining sample. Note that youll also set the DiscretizationMethod and (optionally) the DiscretizationBucketCount properties if you mark your column as Discretized. In our sample, weve set the bucket size to 10 and DiscretizationMethod to Automatic. Possible values for discretization method are automatic, equal areas, or clusters. Automatic means that SSAS determines which method to use. Equal areas results in the input data being divided into partitions of equal size. This method works best with data with regularly
distributed values. Clusters means that SSAS samples the data to produce a result that accounts for clumps of data values. Because of this sampling, Clusters can be used only with numeric input columns. You can use the date, double, long, or text data type with the Discretized content type.
(Smart Business Intelligence Solutions with Microsoft SQL Server 2008, Copyright 2009 by Kevin Goff and Lynn Langit)