Some insights are pretty obviously outliers, so it would be nice to have a confidence level next to insights, based on how many data points the insight is pulled from and possibly the variation between the factor’s effect on all measured days (high variation in effect = less correlation, even if i.e. melatonin causes a 60% average increase in social anxiety)

But I imagine it would be hard to account for variation, since there are so many factors being logged… the more similar the logged days are — besides the factor being measured — the more accurate the data would be, I guess?