• notabot@lemm.ee
    link
    fedilink
    arrow-up
    5
    ·
    2 days ago

    That’s not the only way to do it. In quite a lot of situations you can, instead, generate artificial data that is statistically similar to the original data set and use that instead. That works well for things like system testing, performance tuning and integration testing. Done right, you can even still pull out useful corelations without risking deanonymising the data.