aleixrvr Profile Banner
Aleix Ruiz de Villa Profile
Aleix Ruiz de Villa

@aleixrvr

Followers
904
Following
3K
Statuses
3K

PhD in mathematical analysis. Data scientist. Causal Inference

Vila de Gracia, Barcelona
Joined February 2015
Don't wanna be here? Send us removal request.
@aleixrvr
Aleix Ruiz de Villa
3 years
I'm very happy to announce that my book "Causal Inference for Data Science" has been published at @ManningBooks You will find it in Manning Early Access program, where a new chapter is published regularly every month. (1/4)
1
10
46
@aleixrvr
Aleix Ruiz de Villa
2 months
RT @ManningBooks: 📣 Now in PRINT! 📣 Causal Inference for Data Science, by @aleixrvr 📚Understanding causality hel…
0
1
0
@aleixrvr
Aleix Ruiz de Villa
3 months
@eliasbareinboim Thanks for the wrap up!
0
0
1
@aleixrvr
Aleix Ruiz de Villa
3 months
RT @eliasbareinboim: There is a result in causal inference called the 'causal hierarchy theorem' (Thm. 27.1, , whic…
0
16
0
@aleixrvr
Aleix Ruiz de Villa
3 months
@eliasbareinboim @analisereal @ccaballeroh10 Thanks, that clarifies a lot! Then, I'm re-learning again the basics of of CI 101: you cannot discover the laws of physics without making physical experiments
0
0
0
@aleixrvr
Aleix Ruiz de Villa
3 months
@eliasbareinboim @analisereal @ccaballeroh10 Ah, ok! If I understood correctly, physical laws belong to rung 3 (because you need the explicit functional form of causal relationships)?
1
0
0
@aleixrvr
Aleix Ruiz de Villa
3 months
@eliasbareinboim @analisereal @ccaballeroh10 Hope that clarifies something. Thanks anyway for the help!
0
0
1
@aleixrvr
Aleix Ruiz de Villa
3 months
@eliasbareinboim @ccaballeroh10 @analisereal But performing an intervention is also OOD
0
0
1
@aleixrvr
Aleix Ruiz de Villa
3 months
@ccaballeroh10 @analisereal @eliasbareinboim There is also the work of Jonas Peters et al. about using different environments with the same causal structure.
0
0
1
@aleixrvr
Aleix Ruiz de Villa
3 months
@ccaballeroh10 @analisereal But , bseides intuition, we still need a formal proof that tells us whether we are right or wrong.
@aleixrvr
Aleix Ruiz de Villa
3 months
@analisereal That's a different question. Moreover it is a very interesting one. If you can conclude that you need additional assumptions (for instance causal ones) to make predictions OOD, then it is clear that current supervised learning techinques are limited to predict reality.
1
0
0
@aleixrvr
Aleix Ruiz de Villa
3 months
@ccaballeroh10 @analisereal That's interesting. If you intervene, you are OOD. But is OOD always intervening? That would require a proof, right?
2
0
0
@aleixrvr
Aleix Ruiz de Villa
3 months
@fmg_twtr @analisereal Thanks! I didn't know it. I will take a look!
0
0
0
@aleixrvr
Aleix Ruiz de Villa
3 months
@analisereal And I'm talking about just making predictions OOD, not interventions! (rung one)
0
0
0