Evaluating sample pairs

The sample pairs file(s) contains pairs of members in color-coordinated rows that might be the same member or different members with similar attribute data. Someone familiar with the data, such as a data steward, must review the sample pairs and determine whether the members in each pair are the same or different.

About this task

When this job completes, right-click the successful Generate Threshold Analysis Pairs job that is listed in the Jobs view and select Get job results. The Sample Pair Folder dialog opens, giving you the opportunity to specify the number of files across which to split the sample pairs data. Splitting the data across multiple files enables you to easily divide the evaluation task among multiple data stewards.

Procedure

  1. Click Browse to select the path where you want to save the sample pair file(s). This path can be any drive and directory accessible from the computer running MDM Workbench. The default location is the MDM Workbench workspace folder. To save the file(s) on the operational server, use a mapped network drive to which the MDM Workbench computer user has write access.
  2. In the Base File Name field, type the base file name to use for the sample pair file(s). If the File Count is greater than 1, the base file name is appended with a digit. MDM Workbench appends all sample pair files with the .xls extension.
  3. In the File Count field, select the number of files across which you want to save the sample pair data. The maximum number of files is 25. The default is 1.
  4. Click OK. A confirmation dialog shows the location and filename(s) used to save the sample pair data.

Results

This task can be done using the InfoSphere® MDM Pair Manager application or a third-party spreadsheet application. The InfoSphere MDM Pair Manager provides an easy-to-use interface for selecting match status for the sample pairs. Refer to InfoSphere MDM Pair Manager for instructions on using the application.
Note: The sample pair files might contain sensitive information, such as members' social security numbers. Use the appropriate caution when distributing these files to data stewards for review.