3 ms·
CausalImpact: Causal Inference using Bayesian structural time-series models
- closed 11y agoI love how clean the documentation is. It seems like oftentimes new methods are implemented as an afterthought (if at all). I would need to read the article to know what's it doing under the hood, but felt like it would be very easy to tinker with while I did!
- dj-wonk 11y agoIt is refreshing to see the 'How can I check whether the model assumptions are fulfilled?' section included. Causal modeling is notoriously easy to misinterpret.
- MaxwellM 11y agoAgreed, this documentation is exceptional - It makes something rather complicated look trivially easy to implement. It would great to have a "Analysed with CausalImpact" page like a "built with AngularJs" where we can explore how this was used in the wild. I'm tempted to start applying this immediately, but I look forward to reading http://static.googleusercontent.com/media/research.google.com/en//pubs/archive/41854.pdf http://static.googleusercontent.com/media/research.google.co... before starting.