| | | The fact that the number of infections must be positive allows Bayesian constraints on | | | positivity to improve the precision of their estimate.
>>
>> I have no idea what that means. The 95% confidence interval should have
>> included 0, from the information they gave.
>
> Only if you are a very simple minded frequentist statistician. The world
> has moved on a long way from that era as taught in high schools. >
> Modern Bayesian techniques actually started with Laplace who first used
> it informally as a way to weigh Saturn (and got a surprisingly good
> answer for the time). It got formalised by the Reverend Bayes who gave
> his name to the new branch of statistics and then Jeffries and later
> Jaynes in the twentieth century who formalised the mathematics into
> something that was usable in practice. Your spam filter is probably one
> of the most commonly used applications. It is the basis of most
> astronomical imaging including the picture of the black hole last year.
Indeed. I implemented a Bayesian spam filter in my email client, for example.