recordlinkage

What it is and what it does

RecordLinkage is a Python toolkit for matching records within or across datasets, commonly used for deduplication and entity resolution. It wraps pandas and numpy to provide indexing methods (like blocking and sorted neighbourhood indexing), comparison functions for strings, numbers, and dates, and both supervised and unsupervised classifiers to determine which record pairs are matches. The workflow is: create candidate pairs using an indexer, compute similarity features using a comparator, then classify pairs as matches or non-matches with a classifier like Logistic Regression or ECM.

The package is designed for research and small-to-medium datasets. It integrates directly with pandas DataFrames, making it natural to use in existing data pipelines. Dependencies include scipy and scikit-learn for statistical and machine-learning operations, and jellyfish for string similarity metrics. No known security vulnerabilities are recorded.

Use it for:

• Deduplicate customer records in a CRM by blocking on surname and comparing name, address, and phone fields.
• Link census records across two time periods to track population changes and identify the same individuals.
• Match product catalogs from different suppliers to identify equivalent items for price comparison.
• Resolve duplicate entries in a research database by comparing author names, publication dates, and keywords.
• Merge employee records from multiple HR systems using blocking on department and comparing names and IDs.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

RecordLinkage identifies and matches records within or across datasets using indexing, comparison, and classification algorithms, supporting both deduplication and cross-dataset linking tasks.

Yes, if you need record linkage or deduplication on small-to-medium datasets and can tolerate dormant maintenance. The package is stable, has low install friction, and integrates cleanly with pandas. However, avoid it for production systems requiring active support or if you need compatibility guarantees with very recent versions of numpy, pandas, or scikit-learn.

Install

recordlinkage on PyPI

pip install recordlinkage

uv add recordlinkage

poetry add recordlinkage

Installing recordlinkage

Before you install

Low install friction with a pure-Python wheel and six common dependencies (numpy, pandas, scipy, scikit-learn, joblib, jellyfish). Maintenance is dormant—last release was 2023-07-20 and last commit 2024-02-21, over 1121 days ago—so expect no active bug fixes or feature updates.

License in practice

BSD-3-Clause is permissive and imposes minimal restrictions; you may use, modify, and distribute the package freely in commercial and private projects provided you include the license notice.

Quickstart

import recordlinkage
import pandas

df_a = pandas.DataFrame(data_a)
df_b = pandas.DataFrame(data_b)

indexer = recordlinkage.Index()
indexer.block('surname')
candidate_links = indexer.index(df_a, df_b)

c = recordlinkage.Compare()
c.string('name_a', 'name_b', method='jarowinkler', threshold=0.85)
feature_vectors = c.compute(candidate_links, df_a, df_b)

Requires Python 3.8 or higher.

Verify before relying

• Whether dormant maintenance status affects compatibility with pandas/numpy/scikit-learn versions released after 2024-02-21.
• Performance characteristics on large datasets (the description mentions 'small or medium sized files').

Package facts

Evidence: recordlinkage-0.16-py3-none-any.whl

Tags

More Information Analysis packages