3 ms·
Just to be pedantic: Each of the data points is an average of many users (and potentially have huge errorbars they do not show). Therefore for single individua
by nacc 10y ago
Just to be pedantic:
Each of the data points is an average of many users (and potentially have huge errorbars they do not show). Therefore for single individuals, this result may not be very predictive.
In fact, given that the distribution of number of posts likely follows power law, the average is hugely influenced by a small number of heavy facebook users. I would say this figure summaries more about top facebook users rather than the majority of users. So when "you" (an average facebook user) fall in love, this is probably not what facebook sees.
- jmount 10y agoI thought the same thing (and wrote a bit about it: http://www.win-vector.com/blog/2014/02/the-gap-between-data-mining-and-predictive-models/ http://www.win-vector.com/blog/2014/02/the-gap-between-data-... ).