I assume this may clarify something for those who is already familiar with the topic, but for me its unclear how this is actually predicting anything...

Why, you ask, is it called Benford’s Law when it was discovered earlier by someone else? Stigler’s Law of Eponymy: “No discovery or invention is named after its first discoverer.” In 1983 statistician-historian Stephen Stigler published a paper worthy of Arthur Conan Doyle titled “Who Discovered Bayes’ Theorem?” Since Bayesian analysis, which is at the heart of inductive reasoning, is rapidly becoming the dominant statistical paradigm of the 21st century, this is of considerable scientific interest. Stigler suggests that the Reverend Thomas Bayes (1702-1761) may NOT be the true originator of the theorem named after him. In a tour de force of statistical deduction (or maybe induction) he uses Bayes’ Theorem itself to produce 3 to 1 odds that the real author was Nicholas Saunderson, a blind mathematician who became the fourth Lucasian professor at Cambridge (Newton was the second). He does leave the door open to other possibilities by invoking Damon Runyon’s rule (nothing in life is more than 3 to 1) and Laplace’s principle of indifference (absent contradictory evidence, give equal weight to all possible outcomes).

4:35 💀 only Aric sir can teach in this way Thanks Sir for this series I have learned time series from your channel and it really helped my in my exams

I can't thank him enough for making this series and making it soo much fun

You are a phenomenal educator

You are amazing.. thanks a lot

merci beaucoup monsieur Labarr!

Helpful is your video.

Best presentation ✅👌

Seeing myself learning this in 5 minutes is still a shock Thank you❤

Excellent. Thank you mate

Sooo good. Thanks a lot!

Can we apply arch/garch to arma model? It seems ma part has already modeled the noise/vol. Will adding ma give us more reliable volatility estimates?

Keep up the good work!!!!!!!!! Thanks a lot!

I think there is an error. The "Term" in the fourier series is considered as a pair sin(x)+cos(x) for each n. So if we choose n = 3 (number of terms) we will have 3 pairs of sin(x)+cos(x). @Aric LaBarr

oh..i just stuck gold mine..thanks Arick..

thank u so much sir, you're such a lifesaver!

super useful to understand complex subject. hope to see the rest of machine learning approaches video soon

😀😀😀 @ ""kids that's called marketing!" Thank you so much for this video it helped a lot. You are great at this!

This is what KZbin should be. No fluff, informative, and entertaining. Thank you Doc!

Great vid, many thanks

It sounds like a fancy wrapper over Python's time series decomposition with structural breaks embdedded. It does feel very overfit. Any thoughts on in this ?

A lot of overlap here with an infinite impulse response filter from DSP. Im about to watch the moving average model video, but am wondering if that is the finite impulse response equivalent :)

I like your videos, thanks!

Thanks!

shouldn't it be, if Φ > 1 and not Φ < 1?

@3:23 a lot of bad memory from my calculus days 🤣🤣🤣🤣.

Bro God bless you!

the playlist was amazing. thank you

so clear

so is it possible to read out the algebraic form of the fitted linear model out explicitly?:) If not, is there another approach where this is possible?

Thank you for this excellent video!

I like how you go straight to the point while still making it understandable. Both fast and well-explained.

So, what if there's integration and also seasonal integration?

salute to this dude for the clarity of his explanations

Great and concise! Very engaging especially with the witty titles! Thank you for sharing :)

This video is amazing

nice video!

These were an excellent series, though it would be fantastic if you could do deep dive videos as well! thanks for these regardless

Great informative, funny! Thanks for those videos. I must say, however, some (like this one) are stretching my brain out a bit too thin... some components could do with a little bit more explanation and maybe you got rushed by the 5 minute limit you placed on yourself :)

Best series of videos I've seen on anomaly detection. Great work!

does this mean that we can use bayesian statistics to impose a non-negativity constraint to our statistical models like ARIMA or Holt-Winters to ensure that the forecasted values can never be negative? assuming that it doesn't make sense for the forecasted values to be below zero.

kids thats called marketing.......I subscribed so fast

I was super lost thanks for explaining it amazingly!

Love from Asia ! Topics made easy

Video from 4 years ago, so idk if you read it man, but just so you know, I sincerely consider you a genuinely wonderful person for doing this series

Tellement bien expliquée! merci

3:35 So when I predict for Y_t+1, how will I know the value of error e_t+1? I need it for predicting the value of tomorrow, yet I don't know it, and it's not based on the previous errors.

the underlying assumption is that we know the data up to time t-1, and we use the observed data to estimate the parameters (ϕ1,ϕ2,…,ϕpϕ1​,ϕ2​,…,ϕp​ and e_t) , right?

Just some clarifications. When you say "model the lack of consistency in variance", do you mean model the variance in a consistent way? When you say they are Lazy, do you mean they are using a method that has statistically incorrect properties for the sake of simplicity?