Matching Records from a Single Source - spectrum_quality_1 - 23.1

Spectrum Data Quality Guide

Product type
Software
Portfolio
Verify
Product family
Spectrum
Product
Spectrum > Quality > Spectrum Quality
Version
23.1
Language
English
Product name
Spectrum Data Quality
Title
Spectrum Data Quality Guide
Topic type
How Do I
Overview
Tips
Reference
First publish date
2007
ft:lastEdition
2024-03-04
ft:lastPublication
2024-03-04T22:52:13.486265

This procedure describes how to use an Intraflow Match stage to identify groups of records within a single data source (such as a file or database table) that are related to each other based on the matching criteria you specify. The dataflow groups records into collections and writes the collections to an output file.

  1. In Enterprise Designer, create a new dataflow.
  2. Drag a source stage onto the canvas.
  3. Double-click the source stage and configure it. See the Dataflow Designer's Guide for instructions on configuring source stages.
  4. Drag a Match Key Generator stage onto the canvas and connect it to the source stage.

    For example, if you are using a Read from File source stage, your dataflow would now look like this:

    Read from File in dataflow

    Match Key Generator creates a non-unique key for each record, which can then be used by matching stages to identify groups of potentially duplicate records. Match keys facilitate the matching process by allowing you to group records by match key and then only comparing records within these groups.

  5. Double-click Match Key Generator.
  6. Click Add.
  7. Define the rule to use to generate a match key for each record.
    For more information, see Match Key Generator Options.
  8. When you are done defining the rule click OK.
  9. If you want to add additional match rules, click Add and add them, otherwise click OK when you are done.
  10. Drag an Intraflow Match stage onto the canvas and connect it to the Match Key Generator stage.

    For example, if you are using a Read from File source stage, your dataflow would now look like this:

    Read from File in dataflow
  11. Double-click Intraflow Match.
  12. In the Load match rule field, select one of the predefined match rules which you can either use as-is or modify to suit your needs. If you want to create a new match rule without using one of the predefined match rules as a starting point, click New. You can only have one custom rule in a dataflow.
    Note: Do not use special characters while creating a new rule.
    Note: The Dataflow Options feature in Enterprise Designer enables the match rule to be exposed for configuration at runtime.
  13. In the Group by field, select MatchKey.

    This will place records that have the same match key into a group. The match rule is applied to records within a group to see if there are duplicates. The match key for each record will be generated by the Generate Match Key stage you configured earlier in this procedure.

  14. For information about modifying the other options, see Building a Match Rule.
  15. Click OK to save your Intraflow Match configuration and return to the dataflow canvas.
  16. Drag a sink stage onto the canvas and connect it to the Generate Match key stage.

    For example, if you were using a Write to File sink stage your dataflow would look like this:

    Write to File in dataflow
  17. Double-click the sink stage and configure it.

    For information on configuring sink stages, see the Dataflow Designer's Guide.

You now have a dataflow that will match records from a single source.

Example of Matching Records in a Single Data Source

As a data steward for a credit card company, you want to analyze your customer database and find out which addresses occur multiple times and under what names so that you can minimize the number of duplicate credit card offers sent to the same household.

This example demonstrates how to identify members of the same household by comparing information within a single input file and creating an output file containing one record per household.

Dataflow to create household record

The Read from File stage reads in data that contains both unique records for each household and records that are potentially from the same household. The input file contains names and addresses.

The Match Key Generator creates a match key which is a non-unique key shared by like records that identify records as potential duplicates.

The Intraflow Match stage compares records that have the same match key and marks each record as either a unique record or as one of multiple records for the same household.

The Conditional Router sends records that are collections of records for each household to the Filter stage, which filters out all but one of the records from each household, and sends it on to the Stream Combiner stage. The Conditional Router stage also sends unique records directly to Stream Combiner.

Finally, the Write to File stage creates an output file that contains one record for each household.