Hands-On Microsoft SQL Server 2008 Integration Services part 45. Deploy and manage high-performance data transformation solutions across your enterprise using the step-by-step techniques in this fully revised guide. Hands-On Microsoft SQL Server 2008 Integration Services, Second Edition explains the tools and methods necessary to extract conclusive business intelligence from disparate corporate data. Learn how to build and secure packages, load and cleanse data, establish workflow, and optimize performance. Real-world examples, detailed illustrations, and hands-on exercises are included throughout this practical resource. . | 418 Hands-On Microsoft SQL Server 2008 Integration Services a randomly selected 15 percent of the total records in the segment. One other example could be that you need to create a training data set and a test data set for your data mining models and you want to divide your data set based on a defined percentage. In such a case you can use the Percentage Sampling transformation to create a sample data set from the input rows using the specified percentage. This transformation helps you create a representative data set much smaller in size that you can use for variety of purposes such as testing your packages in a development environment or using the sample data set for surveys and marketing purposes. The Percentage Sampling transformation has a simple and intuitive user interface with just four fields. The first field Percentage Of Rows allows you to specify a percentage of sampling. The Percentage Sampling transformation uses an algorithm to select at random the number of rows according to the specified percentage. However the number of rows that this transformation selects does not precisely match with the percentage calculations . the output rows may be a little bit too many or a little bit too few. The next two fields allow you to specify the names for selected sample output rows and the remaining unselected output rows. As this transformation selects rows for sampling it outputs those rows onto its first output and the remaining unselected rows are outputted on to its second output. This transformation supports one input and two outputs to support both the selected and unselected data sets and supports no error output. The last field is a check box and a value pair you can select the check box if you want to specify a sampling seed and type a value in the field provided as the sampling seed. If you specify a sampling seed and reuse the same sampling seed in a later run it will produce the same sample output no matter how many times you run the package with
