Signature

edf.fix(data, *, strategy='auto', drop_duplicates=True, fill_missing='median', dry_run=False) -> FixResult

Apply automated cleaning fixes to a dataset.

Description

Executes an opinionated cleaning pipeline: normalises whitespace, coerces obvious dtypes, drops exact duplicates, imputes missing values with a per-column strategy, and flags rows that could not be repaired. Returns a FixResult that exposes the cleaned DataFrame, the list of applied fixes and helpers for export.

Parameters

Name	Type	Default	Description
data	str \| pathlib.Path \| pandas.DataFrame	—	Source dataset. Same input types as assess().
strategy	Literal['auto', 'safe', 'aggressive']	'auto'	'safe' only applies non-destructive fixes; 'aggressive' will drop columns with >90% missing values.
drop_duplicates	bool	True	Remove exact duplicate rows before imputation.
fill_missing	Literal['mean', 'median', 'mode', 'ffill', 'drop']	'median'	Per-column imputation strategy for numeric fields.
dry_run	bool	False	When True, computes the diff without materialising the cleaned DataFrame.

Returns

FixResult — Wraps .dataframe, .applied_fixes, .to_csv(), .to_excel() and .diff().

Raises

ValueError — an unknown strategy or fill_missing option is passed.

Examples

Clean and export

fix_example.py

python

Colab GitHub

import eazydatafix as edf

result = edf.fix("employees.csv")
result.applied_fixes
result.to_csv("clean.csv")

Python 3.11— Expected output

>>> result = edf.fix("employees.csv")
>>> result.applied_fixes
['strip_whitespace', 'coerce_numeric', 'drop_duplicates(6)', 'impute_missing(38, median)']
>>> result.to_csv("clean.csv")
wrote clean.csv (1,198 rows × 12 cols)

Notes

fix() is deterministic — the same input plus the same options produces the same output.
Pass dry_run=True to preview which fixes would apply without touching the data.

Best Practices

Store result.applied_fixes alongside the cleaned file so pipelines stay auditable.
Prefer strategy='safe' for regulated workloads; keep 'aggressive' for exploratory work.

edf.fix()