4 ms·
Precisely (or to be really specific, starting with a suspicion of causation and using correlation as one piece of weak evidence).
by asdfasgasdgasdg 5y ago
Precisely (or to be really specific, starting with a suspicion of causation and using correlation as one piece of weak evidence).
- ahmedelsama 5y agoI think the ideal is start with something that you think is causal, then do the discussed correlation analysis then it it looks fruitful, start testing by changing the distribution of the feature and if it is causal then you should see the kpi move in the same direction. Does that make sense? Smaller steps to make sure you only invest if it is worth it