levenshtein_at_thresholds("Description",2)

[msiemionCalistapw]

what does the ",2" and ",[1,2]" mean in levenshtein_at_thresholds ?

[RobinL]

That will create one comparison level with a threshold distance of 1, and a second at a threshold distance of 2

There's some documentation here:

splink/splink/comparison_library.py

Line 153 in 5af45c5

An example of the output with default arguments and setting

And the comparison has a human_readable_description property that will explain what it does:

splink/splink/comparison.py

Line 467 in 5af45c5

def human_readable_description(self):

So for instance:

from splink.duckdb.duckdb_comparison_library import (
    levenshtein_at_thresholds,
)

print(levenshtein_at_thresholds("my_col", [1,2,3]).human_readable_description)

will output

Comparison 'Exact match vs. levenshtein at thresholds 1, 2, 3 vs. anything else' of my_col.
Similarity is assessed using the following ComparisonLevels:
    - 'Null' with SQL rule: my_col_l IS NULL OR my_col_r IS NULL
    - 'Exact match' with SQL rule: my_col_l = my_col_r
    - 'levenshtein <= 1' with SQL rule: levenshtein(my_col_l, my_col_r) <= 1
    - 'levenshtein <= 2' with SQL rule: levenshtein(my_col_l, my_col_r) <= 2
    - 'levenshtein <= 3' with SQL rule: levenshtein(my_col_l, my_col_r) <= 3
    - 'All other comparisons' with SQL rule: ELSE