LoginSign Up
← Back to all sparks
C

cleanepi

ANALYTICS
Velocity0.0

R package to clean and standardize epidemiological data

epiverse-trace.github.io/cleanepi ↗

cleanepi is in the long tail of bug fixes that follows a 1.0 — and changed maintainers along the way.

epiverse-tracedata-cleaningline-listreportingepidemiology
◆Current state
cleanepi cleans and standardises epidemiological line list data — dates, subject IDs, missing values, duplicates — and produces a report of what it changed. Since 1.0.0 in mid-2024 the releases have been almost entirely corrective: date-guesser fixes, report structure fixes and matching behaviour corrections. Maintainership passed to Bubacarr Bah in 1.1.2.
◆Where it's heading
Work has concentrated on the report object and on making the cleaning functions behave predictably at the edges — case- and whitespace-insensitive missing-value matching, report elements returned as vectors instead of comma-separated strings, an argument to print a single operation's report. The underlying cleaning API has barely moved since 1.0.0, which suggests it is settled.
◆Prediction
The report interface has been reworked repeatedly across these releases and is the most likely place for further change; the cleaning functions themselves look stable.

◆Recent moves

  1. 9mo ago

    Case-insensitive missing-value matching and report fixes

    replace_missing_values() now matches na_strings ignoring case and surrounding whitespace — a behaviour change for anyone relying on exact matching — alongside cli-related errors fixed in check_subject_ids() and duplicate elements removed from the report object.

    View source ↗
  2. 1y ago

    Report elements become vectors; date parsing default restored

    Several report elements that were comma-separated strings are returned as vectors, print_report() gains a what argument to print one operation's report, and the standardize_dates() orders default from 1.0.2 is reinstated.

    View source ↗
  3. 1y ago

    Date guesser corrected; empty-row indices fixed

    The date guesser handles quarters and partial dates and pre-processes odd cases before {lubridate} parses them, print_report() returns actual empty-row indices rather than a row count, and automatic numeric-to-date conversion is stopped because it produced unexpected results.

    View source ↗
  4. 2y ago

    First major release with cleaning and reporting

    1.0.0 establishes the two halves of the package that everything since has refined: the data cleaning operations and the report describing what they did.

    View source ↗