News for hackers. Stuff that matters.
Re:Han Solo and Bayesian Priors by p1mrx on Saturday March 28, 2015 @07:07PM (#9283135)
You also need to consider the survivorship bias of a story from a long time ago in a galaxy far, far away.

All the other potential stories where the protagonist predictably wins the Darwin Award would be too uninteresting to reach us over such an immense time and distance.

Re:Han Solo and Bayesian Priors by akozak on Saturday March 28, 2015 @12:37PM (#9281938)
C3PO's estimation would have been a lot more useful if he'd calculated their odds.

Also his calculation suggests there's a strong open data regime in the empire, which is a nice thought.

Re:Han Solo and Bayesian Priors by aet on Saturday March 28, 2015 @12:49PM (#9281969)
The odds may have been based on simulation rather than "real data"
Re:Han Solo and Bayesian Priors by akozak on Saturday March 28, 2015 @01:16PM (#9282053)
Ah but in that case I find it hard to believe a robot wouldn't take piloting skills into account.

For example when wind energy developers study wildlife impact they tend to include avian "avoidance behavior" in modeling risk of collision.

Re:Han Solo and Bayesian Priors by ced on Saturday March 28, 2015 @01:10PM (#9282039)
I like the concept, it's a cool way of introducing Bayesian probabilities!

The conclusion strikes me as overly confident, though. It implies that if we had 100 Han Solo's, we are very confident that about 3/4 of them are going to make it through. This comes about because it uses this:

We're going to say that C3PO has records of two people surviving and 7,440 people ending their trip through the asteroid field in a glorious explosion.

as data for Han's odds of making it. But 100 out-of-shape people dying on the ascent of Mount Everest does not tell us anything about the odds for someone who is very fit.

I would rather model that there is no one true probability of making it through - it's going to be dependent on the pilot. Mediocre pilots might have odds in the neighborhoud of 1/3720, but there is presumably a lot of variance depending on skill, and my prior belief is that Han would wind up in the upper end of the distribution.

Re:Han Solo and Bayesian Priors by sukilot on Saturday March 28, 2015 @05:31PM (#9282829)
How is your model different from the blog post? Yours sounds like a paraphrase.

Do you disagree that Han is more likely to survive more difficult challenges than easier ones?

Re:Han Solo and Bayesian Priors by ced on Saturday March 28, 2015 @05:52PM (#9282874)
The blog post has the right idea, but the wrong equations. If we take C3PO's statement to mean:

We're going to say that C3PO has records of two people surviving and 7,440 people ending their trip through the asteroid field in a glorious explosion.

Then the big question is how skill-dependent the challenge is. Is it like swimming across the English channel, or is it random like surviving the Niagara Falls in a barrel? There exists a distribution of survival rates as a function of skill level, we just don't know what it is. We could assume a sigmoid form. C3PO's statement gives us some constraint on what the parameters of the sigmoid are, but not enough. We could make some further assumptions and get an answer, but the results will depend strongly on the specific assumptions. In other words: we don't know the answer. https://xkcd.com/384/

Re:Han Solo and Bayesian Priors by xaetium on Saturday March 28, 2015 @09:12PM (#9283494)
I often find myself wondering if the major problem with any given statistical analysis is whether Bayesian inference ought to have been used at all.
Re:Han Solo and Bayesian Priors by Matumio on Saturday March 28, 2015 @02:18PM (#9282243)
The 75% result (chance to survive) seems about right, but the confidence in this result seems completely unnatural. The sharp peak in the posterior plot suggests that I have nailed down the exact probability to +/- 2%. No way.
Re:Han Solo and Bayesian Priors by brobo on Saturday March 28, 2015 @01:43PM (#9282146)
Thought experiment: imagine there is a second robot, C4P1, on one of the TIE fighters. He independently comes up with an estimate of 10,000:1 for Han's death. Are we now going to slide our estimate all the way down? Or, what if C3P0's line was revised to say 1,000,000:1? Would the author now say, 'whelp, I was pretty confident at 20,000:1, but now that C3PO's number is so much bigger than mine, I guess Han's gonna die'?
Re:Han Solo and Bayesian Priors by EGreg on Sunday March 29, 2015 @02:37AM (#9284098)
I think this is the kind of analysis you need instead:

http://tvtropes.org/pmwiki/pmwiki.php/Main/PlotArmor

Re:Han Solo and Bayesian Priors by mikexstudios on Saturday March 28, 2015 @04:08PM (#9282592)
Re:Han Solo and Bayesian Priors by elliott34 on Saturday March 28, 2015 @02:09PM (#9282215)
Does the author have something backwards or do I?

P(RateOfSuccess|Successes) = Beta(α,β)

"In Bayesian terms, C3PO's estimate of the true rate of success given observed data is referred to as the likelihood."

BUT we know likelihood(rateofsuccess|data) = probability(data|rate of success).

I am confused.

Re:Han Solo and Bayesian Priors by jmcohen on Saturday March 28, 2015 @05:15PM (#9282787)
"Likelihood" can refer to both p(param|data) and p(data|param). See [1] for an example of the first, and [2] for the second.

[1] http://en.wikipedia.org/wiki/Likelihood_function

[2] http://en.wikipedia.org/wiki/Bayes%27_theorem#Events

Re:Han Solo and Bayesian Priors by UhUhUhUh on Saturday March 28, 2015 @03:45PM (#9282529)
One factor seems missing: the deviation of Han's skills from that of the average pilot. Does C3PO take it into account in his likelihood? Do we take it into account in our probability of success? Although I do get the point, it seems there's often something of that sort, a simplification, in that kind of bayesian reasoning.
Re:Han Solo and Bayesian Priors by lordnacho on Saturday March 28, 2015 @01:53PM (#9282175)
Someone needs to do this with Game of Thrones characters. Everything I thought about important characters' survival chances turned upside down as I watched it.
Re:Han Solo and Bayesian Priors by davmre on Saturday March 28, 2015 @02:39PM (#9282306)