Depends on how many cases of "large scale events" you have. You could calculate a simple correlation coefficient and even a p-value to loosely estimate causation, if you had enough data points.
As a noob in statistical analysis I'm wondering how would one calculate the correlation coefficient between two sets of dates (occurrences of something happening, in this example outages/SEV1s and days of Mercury retrograde)?
I guess the grand irony of all this is that data scientists — of which, the "good" ones (rational Spock-types, highly technical, sometimes to their own detriment) are, historically, not socially inclined — are the ones responsible for optimizing _social_ media platforms in the first place.
But yeah, something something Mercury.
I think what you could do (scientists please feel free to correct me) is you basically separate days into two groups, 1/ days falling during mercury retro 2/ days falling during not-mercury retro (I think called direct motion). Then you look at the total LSEs/total days for each of the two groups and do a two-sample t-test...? Depending on the p-value you can conclude there is some correlation. Now for the fun part.. data acquisition. Anyone know a good source for tabulating LSEs/day?