make_deduplication_data#

skrub.datasets.make_deduplication_data(examples, entries_per_example, prob_mistake_per_letter=0.2, random_state=None)[source]#

Duplicates examples with spelling mistakes.

Characters are misspelled with probability prob_mistake_per_letter.

Parameters:
exampleslist of str

Examples to duplicate.

entries_per_examplelist of int

Number of duplications per example.

prob_mistake_per_letterfloat in [0, 1], default=0.2

Probability of misspelling a character in duplications. By default, 1/5 of the characters will be misspelled.

random_stateint, RandomState instance, optional

Determines random number generation for dataset noise. Pass an int for reproducible output across multiple function calls.

Returns:
list of str

List of duplicated examples with spelling mistakes.

Examples

>>> from skrub.datasets import make_deduplication_data
>>> make_deduplication_data(["string1", "string2"],
...                         entries_per_example=[4, 5],
...                         random_state=9)
['btrwng1', 'string1',
'string1', 'string1',
'saoing2', 'string2',
'string2', 'string2',
'string2']