Intro

Recently I’ve come to the link [1] and it’s quite interesting and uncommon – just for funny knowledge though, nothing ext scientific.

Correlation vs Causation

I remember studying statistics at bachelor/masters and the correlation was an interesting topic for me, really.  How come two things could be so correlated but not cause the other?! But yeah. Correlation does not imply causation. The examples [1] are there to show.

My thesis

{Reminder: my thesis was about measuring performance metrics and execution of C++ programs}

Interestingly one of the bases of my master thesis is the assumption that correlation implies causation – which is quite a strong statement, but for the specific cases we deal and assuming the metrics measure all the system, we can then assume causationXcorrelation (X== implies). This is possible using the causal modal I assumed for the system [2] by the common cause principle of Reichbach [3].
It will be clearer below:

However, you can infer causality from correlation

Using a causal network, with 20 possibilities to know causality, in the model below (stolen from [2]) we assume then: (N ->) Abnormality -> Metric. The model is very useful and should it will be used for other analysis. The one I considered on the models for the metrics.

REFs

[1] http://www.tylervigen.com/spurious-correlations?fbclid=IwAR0Uz8MueFHvF994n3_sMFBzQY3IAlkUIRcuqfbyisL4doTWkzOszc3d7jM

[2] https://www.youtube.com/watch?v=HUti6vGctQM

[3] https://plato.stanford.edu/entries/physics-Rpcc/