The Dedupe library made easy with Pandas.

pip install pandas-dedupe

Basic Deduplication

A training file and a settings file will be created while running Dedupe. Keeping these files will eliminate the need to retrain your model in the future.

If you would like to retrain your model from scratch, just delete the settings and training files.

dedupe_dataframe is for deduplication when you have data that can contain multiple records that can all refer to the same entity

import pandas as pd
import pandas_dedupe

#load dataframe
df = pd.read_csv('test_names.csv')

#initiate deduplication
df_final = pandas_dedupe.dedupe_dataframe(df,['first_name', 'last_name', 'middle_initial'])

#send output to csv
df_final.to_csv('deduplication_output.csv')

gazetteer_dataframe is for matching a messy dataset against a 'canonical dataset' (i.e. the gazette)

import pandas as pd
import pandas_dedupe

#load dataframe
df_clean = pd.read_csv('gazette.csv')
df_messy = pd.read_csv('test_names.csv')

#initiate deduplication
df_final = pandas_dedupe.gazetteer_dataframe(df_clean, df_messy, 'fullname', canonicalize=True)

#send output to csv
df_final.to_csv('gazetteer_deduplication_output.csv')

Use identical field names when linking dataframes. Record linkage should only be used on dataframes that have been deduplicated.

import pandas as pd
import pandas_dedupe

#load dataframes
dfa = pd.read_csv('file_a.csv')
dfb = pd.read_csv('file_b.csv')

#initiate matching
df_final = pandas_dedupe.link_dataframes(dfa, dfb, ['field_1', 'field_2', 'field_3', 'field_4'])

#send output to csv
df_final.to_csv('linkage_output.csv')

The canonicalize parameter will standardize names in a given cluster. Original fields are also kept.

pandas_dedupe.dedupe_dataframe(df,['first_name', 'last_name', 'payment_type'], canonicalize=True)

Group records into clusters only if the cophenetic similarity of the cluster is greater than the threshold.

pandas_dedupe.dedupe_dataframe(df, ['first_name', 'last_name'], threshold=.7)

If True, it allows a user to update the existing model.

pandas_dedupe.dedupe_dataframe(df, ['first_name', 'last_name'], update_model=True)

Specifies the sample size used for training as a float from 0 to 1. By default it is 30% (0.3) of our data.

pandas_dedupe.dedupe_dataframe(df, ['first_name', 'last_name'], sample_size=0.5)

If you'd like to specify dates, spatial data, etc, do so here. The structure must be like so: ('field', 'type', 'additional_parameter). the additional_parameter section can be omitted. The default type is String.

See the full list of types below.

# Price Example
pandas_dedupe.dedupe_dataframe(df,['first_name', 'last_name', ('salary', 'Price')])

# has missing Example
pandas_dedupe.link_dataframes(df,['SSN', ('bio_pgraph', 'Text'), ('salary', 'Price', 'has missing')])

# crf Example
pandas_dedupe.dedupe_dataframe(df,[('first_name', 'String', 'crf'), 'last_name', (m_initial, 'Exact')])

Dedupe supports a variety of datatypes; a full list with documentation can be found here.

pandas-dedupe officially supports the following datatypes:

• String - Standard string comparison using string distance metric. This is the default type.
• Text - Comparison for sentences or paragraphs of text. Uses cosine similarity metric.
• Price - For comparing positive, non zero numerical values.
• DateTime - For comparing dates.
• LatLong - (39.990334, 70.012) will not match to (40.01, 69.98) using a string distance metric, even though the points are in a geographically similar location. The LatLong type resolves this by calculating the haversine distance between compared coordinates. LatLong requires the field to be in the format (Lat, Long). The value can be a string, a tuple containing two strings, a tuple containing two floats, or a tuple containing two integers. If the format is not able to be processed, you will get a traceback.
• Exact - Tests whether fields are an exact match.
• Exists - Sometimes, the presence or absence of data can be useful in predicting a match. The Exists type tests for whether both, one, or neither of fields are null.

Additional supported parameters are:

• has missing - Can be used if one of your data fields contains null values
• crf - Use conditional random fields for comparisons rather than distance metric. May be more accurate in some cases, but runs much slower. Works with String and ShortString types.

Tyler Marrs - Refactored code, added docstrings, added threshold parameter

Tawni Marrs - refactored code, added docstrings

ieriii - Added update_model parameter, updated codebase to use Dedupe 2.0, added support for multiprocessing, added gazetteer_dataframe.

Daniel Marczin - Extensive updates to documentation to enhance readability.

Alexis-Evelyn - Fixed logger warning with related to Pandas.

Niels Horn - Cleaned up utility functions.

Many thanks to folks at DataMade for making the the Dedupe library publicly available. People interested in a code-free implementation of the dedupe library can find a link here: Dedupe.io.