I wonder if it would be appropriate, even if useless, to distinguish between correlations that are utterly unrelated vs benign third factors. For example, household spending on eggs would go up over time with inflation, and Emerson Electric stock price would (might?0 go up (if there is no splitting) over time for the same reason. That is a spurious correlation because the processes are unrelated or not causal of each other, but the numbers follow the same baseline change; if you detrended both data sets and replotted you would see the randomness.
But something like "I can't even" and yogurt prices don't have an underlying baseline that is shifting in the same direction over time.
I especially love the use of two y-axes to make sure that the graphs line up so nicely.
Not knowing this is how superstitions are born.
In the management world, it ends up being the reason for stupid processes and rituals.
Indeed, superficial determination of causation leads to far too many processes that simply do not work.
Thanks for sharing your perspective, Suresh!
Great project.
I wonder if it would be appropriate, even if useless, to distinguish between correlations that are utterly unrelated vs benign third factors. For example, household spending on eggs would go up over time with inflation, and Emerson Electric stock price would (might?0 go up (if there is no splitting) over time for the same reason. That is a spurious correlation because the processes are unrelated or not causal of each other, but the numbers follow the same baseline change; if you detrended both data sets and replotted you would see the randomness.
But something like "I can't even" and yogurt prices don't have an underlying baseline that is shifting in the same direction over time.
I especially love the use of two y-axes to make sure that the graphs line up so nicely.
Great point! Thanks, Greg.
Correlations with a hidden common cause are different from pure coincidence. That's worth exploring. Thanks for the thoughtful comment!