flâneur

Link type - linking vs deduping - Splink

moj-analytical-services.github.io · 281 words · saved by 1 readers

Linking refers to finding links between datasets, whereas deduplication finding links within datasets. Data linking is therefore only meaningful when more than one dataset is provided. This guide shows how to specify the settings dictionary and initialise the linker for the three link types. The dedupe_only link type expects the user to provide a single input table, and is specified as follows The link_only link type expects the user to provide a list of input tables, and is specified as follows: The input_table_aliases argument is optional and are used to label the tables in the outputs. If not provided, defaults will be automatically chosen by Splink. The link_and_dedupe link type expects the user to provide a list of input tables, and is specified as follows: The input_table_aliases argument is optional and are used to label the tables in the outputs. If not provided, defaults will be automatically chosen by Splink.

Link type: Linking, Deduping or Both ¶ Splink allows data to be linked, deduplicated or both. Linking refers to finding links between datasets, whereas deduplication finding links within datasets. Data linking is therefore only meaningful when more than one dataset is provided. This guide shows how to specify the settings dictionary and initialise the linker for the three link types. Deduplication ¶ The dedupe_only link type expects the user to provide a single input table, and is specified as follows from splink import SettingsCreator settings = SettingsCreator ( link_type = "dedupe_only" , )

related reading