This PR aims to fix issue #37 by splitting up a large predictions array into smaller ones, calculating the personalization score for each sample, and then returning the median or the mean of those scores. The default parameter n_too_large is set to 1000; smaller arrays will be treated normally. Let me know what you think @statisticianinstilettos
This PR aims to fix issue #37 by splitting up a large predictions array into smaller ones, calculating the personalization score for each sample, and then returning the median or the mean of those scores. The default parameter
n_too_large
is set to 1000; smaller arrays will be treated normally. Let me know what you think @statisticianinstilettos