Видео

How can I buy a lottery ticket be a hypothesis? 27:03

I’m calling the cops 👮‍♀️ 😂

I’m not your student 🤨

I see why yall hate yall only got 139 views 😂😂😂😂

Why yall stealing

Powerful, thanks for sharing ❤

This is an excellent introduction to Bayesian NMA. Thank you.

Promo'SM 😇

i can't hear anything. what a horrible presentation

This is a good distillation of the ideas of E.T. Jaynes, however, I would welcome further distillation in the form of interactive (graphics??) exercises involving geometric invariances. I don't mean in the form of Venn diagrams but something like a bivector whose area does not depend on a boundary shape such as a circle, square or triangle. R. Buckminster pointed out many years ago that raising a number to the second power can just as easily be represented with a triangle as a square. So I would like to be able to move a slider controlling an area of 100 unit shapes shown as either triangles or squares. I don't know if this is a good idea or not but I would like something more visual and interactive to work with. Years ago I was involved in a question about randomizing the first replicate of entries in an agricultural experiment and realized that if each entry is assigned a colour rather than a number, it didn't matter if I called the entry in plot 1, entry #1 or entry #5 because all the blue entries would be in the same places in the other replicates regardless of whether the blue entry was called entry #1 or entry #5. I would also like to see the transcript of an instructional video (to accompany the interactive WALKME-type interactive exercises) indexed conceptually with categorized question-based indexing as used in ASK hypermedia systems. Chapter titles can be very uninformative compared to categorized and timestamped hyperlinked text questions in a conversational reading system. ??? 38:48 "six out of ten" but there are 11 dots. 7 tan and 4 red.

19:45 "Probability is a numerical repreesntation of logical plausibility, conveniently rescaled". Such an elegant description! I came here after reading "Probability Theory: The Logic of Science" because I wasn't able to articulate the main message of what I read. You have done this beautifully.

I wonder if there is a much deeper problem. For many types of dynamics and problem spaces, P[*], the idea of simple probability, is just not very coherent way of saying anything useful. Crushing a system down into a single dimension, trying to shoehorn in poorly understood measurements and data and poorly understood models for 'P', not to mention relying on all sorts of foundational assumptions about the structure of cause/effect, unfounded assumptions about randomness, and statistics just doesn't make sense in many cases, like discussing medicine in terms of the 4 humors. Bayesian thinking doesn't save you, it just lets you recognize the even deeper problem, it's not just not that P[E | H] != P[E | H], the problem is there is no reasonable way to know if E or H are even sensical concepts that capture the dynamics you intend, if their relationship is meaningful, and no way to know if taking 'P' of them maps to anything useful or is subtly misleading in ways you cannot imagine.

This was very interesting. Thank you!

Cox theorems as Cox wrote them (and as Jaynes repeated) were found later to not be rigorous and have a flaw in which Halpern and others found counterexamples. For example, it disallows belief that only has finite gradiations. To patch Cox Theorem to make a Cox-like Theorem, you need to introduce things that do not make it natural. Kolmogorov's axioms do not suffer from this- they are rigorous as stated and quite intuitive. IMO "probability" is not the same as belief/chance/uncertainty. If are indifferent to A, that does not make P(A) anything that we assign it, it just means we are indifferent. If there are n choices and you make P(A)=1/n, you have not measured the probability of A by any repeated well-designed experiment. Logic IMO cannot be the basis of probability, -only repeated well-designed experiments that frequentism gives a nod to can be, since logic often fails to correspond with reality. For example, what is 7 Bigfoot + 2 Bigfoot? Logic says 9 Bigfoot, but cannot be since Bigfoot doesn't exist. There is no support for any proposition there except in a mathematical fantasy world.

it is a great resource to learn from the pkg author. thank u all for all the great work!

Great talk, really clear-reasoning, very didactic and especially usefull from the perspective of social sciences, thanks

Read my review here statisticool.com/mathstat/claytonsfallacy.html The TLDR version of my review is that Clayton oversells Bayes and undersells frequentism.

Thanks for a great video and nice questions to Will Landau