finds.dev← search

// the find

rhiever/datacleaner

★ 1,082 · Python · MIT · updated May 2019

A Python tool that automatically cleans data sets and readies them for analysis.

datacleaner is a thin wrapper around pandas and scikit-learn that automates the most mechanical parts of tabular data prep: dropping or imputing NaNs, and label-encoding categoricals. It targets data scientists who work with already-structured CSVs and want those two steps done without writing the boilerplate themselves.

The autoclean_cv function is the one genuinely useful idea here — it fits encoders only on the training set and applies them to the test set, which is exactly the mistake beginners make and a real source of leakage. The CLI interface is a nice touch for one-off cleaning without opening a notebook. The API surface is minimal, which means there is almost nothing to learn.

Last pushed in 2019, targeting Python 2.7 and 3.5 — both end-of-life. The imputation strategy is hardcoded median/mode with no way to override, which is fine for demos but breaks on any real dataset where that default is wrong. The feature set is so narrow (drop NaNs, impute, label-encode) that pandas itself plus two lines of sklearn code does the same thing without the dependency. The Travis CI badge links to a defunct service and the Landscape code-health badge is also dead, so you can't tell if the test suite even passes anymore.

View on GitHub →

// want more like this?

We dig through GitHub every week and send a few repos picked for what you actually care about — each with an honest take like this one.

Get finds in your inbox → Search again →