3 ms·
With a ton of data like most start-ups have the effect of the priors generally wash away. That is the likelihood (of the data) generally dominates. My results
by RA_Fisher 13y ago
With a ton of data like most start-ups have the effect of the priors generally wash away. That is the likelihood (of the data) generally dominates. My results are generally very robust to choice of prior parameters (when the type of distribution is well known like Beta for Bernoulli RVs).
- diminoten 13y agoI have literally no idea what you wrote here. Start-ups? Priors? Beta for Bernoulli RVs? "That is the likelihood generally dominates"? "My results are generally very robust to choice of prior parameters"? Neither of those sentences are comprehensible.
- yummyfajitas 13y agoI have a blog post from a while back which explains in more detail everything in RA_Fisher's comment: http://www.chrisstucchio.com/blog/2013/bayesian_analysis_conversion_rates.html http://www.chrisstucchio.com/blog/2013/bayesian_analysis_con...
- diminoten 13y agoYeah, this is mathematical application of bayesian analysis, which is a completely valid and proven way to do analysis of actual data. I'm talking more to the folks over on LW who say things such as, "It's 17% likely that my actions will cause harm, so I'll act./won't act."
- yummyfajitas 13y agoWell, to explain the phrase "My results are generally very robust to choice of prior parameters"?, just look at my section on what happens if you pick different priors. In the example I give, the startup guy and the VC choose different priors but the posteriors are nearly identical. I believe that's what RA_Fisher was describing. That exact situation happened to me and my cofounder a few days ago. I'll be blogging it soon.