4 ms·

Thanks for the article! I love methods :-)

• uniqueuid 5y ago
  Awesome! Maybe you'll like this better method then: Bayesian counterfactual time-series analysis [1]. In essence, it compares a "neutral" time series and another one affected by a natural experiment and produces bayesian point-estimates for the causal impact. Super powerful, there's a great R package, and really easy to use. [1] https://research.google/pubs/pub41854/ https://research.google/pubs/pub41854/
  • ContrarianBrit 5y ago
    Thanks! I'll check it out :-)