Sampling Results

The Sampling Results option gives you the ability to import your class assignments after you have assigned a fingerprint to your class. The class assignments are based on confidence values calculated using threshold recognition algorithms to determine the similarity of contents.

For this Import Type, you need to provide your workspace information (machine ID and company) and Content Threshold values. If desired, you can select other data store information to further restrict the class assignments. Select Import to assign classes to the data elements that have a fingerprint that matches the prototype fingerprint to the degree specified by the Content Threshold values.

Import Class - Sampling Results

For both the Distributed Systems and z/OS solutions, the Import Class window is the similar when Sampling Results is selected. However, it is only appropriate to import your classes in this way if your data stores are distributed (it works both with DB2 on MVS knowledge base and with XDB knowledge base).

Here is a summary of the parameters used by this function:

Machine ID - Identifier of the machine containing the data stores where Import Class process will work.
Company - Identifier of the company containing the data stores where Import Class process will work.
Data Store Qualification - With this set of selection parameters, it is possible to specify the scope of the class assignment process (which data stores you want to assign to the specified class).
Selected class - By using the Add botton it is possible to select one or more classes for the assignment process. This mean the process will attempt the assignation of the selected class or classes.
Super Classes - Selects a super class. This mean the process will attempt the assignment of the selected superclass.
Thresholds - Logical operations between Treshold1 and Treshold2. Data Express calculates two distinct formulas in order to understand how the processed column data distribution fits with the one of the prototype (by comparing compressed sampling result). Each one of the two formulas returns a number in the range 0-100, that is as higher as the two distributions are similar. By setting the two thresholds, and by setting the and/or parameter, it is possible to guide Data Express in the class assignment based on comparison between prototype and column to be assigned.

If you are working with z/OS data stores and a mainframe KB, you must schedule a series of jobs to perform the sampling functionality for you. For instance, you would schedule the job Data Store Data Element Sampling to select the sampling configuration, schedule the job Class Data Element Assignment to suggest class assignments, and then confirm the class assignments in the Work with Data Elements window of Data Builder.

For more information about assigning classes for your z/OS data stores, see section Data Analysis of chapter Project Implementation in the Process for z/OS Guide

For more information about importing classes by sampling results for distributed data stores, see the Simple Sampling exercise in this guide.