@viewermetrics why 50% and not 30%/60% as half baseline? just random guess?
10% as bot limit also kinda random + you may classify a lot of accounts as bots with your method
@viewermetrics pre-2020 analysis instead of a last time period (year?, 2 years?)
not working with median instead of max value lol
random 0.5 value instead of z-score or something like that
every account is a bot inside spike periods, what about special events, is everyone a bot?