4 ms·
I'm in the middle of it and I feel - maybe weird could be a right word, regading the book. It is very verbose in explaining how stupid statistics and statistic
by beefield 7y ago
I'm in the middle of it and I feel - maybe weird could be a right word, regading the book.
It is very verbose in explaining how stupid statistics and statisticians were before the causal revolution orchestrated mainly by the author were. And how magnificiently efficient and simple these new concepts with causation diagrams are for uncovering causal relationships.
But for some reason I have completely missed how you come up with these diagrams in the first place, and how you actually, practically use the data to validate if the diagram you have come up with is correct. In other words, the book has completely failed to help me build any kind of mental model how I should apply this magnificent new idea in practice.
Is that just me being dumb or would there be some other sources worth reading for yhe same concepts? (Yes, I am intrigued with the question how to evaluate causal statements)
- merlinsbrain 7y agoI think you’ll enjoy this piece titled “Bayesian Networks without Tears”. https://www.cs.ubc.ca/~murphyk/Bayes/Charniak_91.pdf https://www.cs.ubc.ca/~murphyk/Bayes/Charniak_91.pdf The author of this paper/article attempts to make Judea Pearl’s concepts digestible while keeping a good balance between using math (not much) but still referring to the relevant mathematical concepts. I’m currently (re-)exploring these concepts (specifically Bayesian networks) and would be happy to chat (check profile) if you end up reading this paper or finding different sources which help you grok the field!