Matching/Merging

Matching and merging refers to the process of linking and combining records in different datasets. As an example, a company may have business customer records organized by account number, and would like to merge the dataset with secondary information available from a general list of businesses. The general business list will be organized by business name, with standardized name and address formats. Customer information is often organized by account number and will also contain business name, address, and location information. In the matching and merging process, the two lists are joined.

Matching typically employs one of two approaches. In a simple exact match, records in two different datasets are deemed to match if the linking or “decision” variable is identical. Simple exact matching is trivial but often under-represents the number of linkages in datasets because name and address identifiers are collected in many different formats. Probabilistic matching is used to identify and link records from one dataset to another on the basis of a calculated statistical probability. Probabilistic matching relaxes the rigid quality of exact matching, replacing the exact match with a fuzzy probabilistic match whose statistical properties are controlled by the matching analyst. 

In National Analysts’ experience, all probabilistic matching requires that questionable cases be reviewed manually to increase match rates and reduce errors. National Analysts has experience matching very large datasets (12 million+ records) and fine-tuning matching algorithms to obtain a very high proportion of matches and very few errors.

Matching/Merging
Fusion
Scoring

For more information on National Analysts, feel free to e-mail us today or call (215) 496-6800.