This is an absolutely great channel. i was reading page rank in the university and thought i understood markhov chains well, but upon watching your video i realised there was so much more intuitive things to learn, Thanks a lot!!!

Absolutely great, useful and easy to follow. Many thanks!

Excellent! The eigenvector perspective is cool - shows how they, once again, reveal the most important characteristics of a system.

@NormalizedNerd

3 жыл бұрын

Yeah...exactly!

Great Video! Thank you

Lol just solved this problem for my code right before you posted this video! Your explanation would have saved me a decent amount of time

@NormalizedNerd

3 жыл бұрын

Haha...it happens XD

Man your outro is awesome!

Great video! I have a doubt. I think the probability function should be divided by "len (seq)", do I?

Just a question. If I would like to find outlier sequences, say, the ones with a very low probability, it is clear that I just need to use your last example and calculate some kind of percentile threshold on my dataset of probabilities of chains to isolate the outliers. I am dealing with sequences of different length, so, bigger sequences will naturally tend to 0 due to the fact that probabilities are between 0 and 1. The question is... I am using geometric mean of the transitions, this is, after multiplying all the probabilities like you do, I calculate the nth root of that product, this to get a "better value" that is not affected by the chain length (in this case n where n can be between "a" and b" for some a,b) in order to be able to compare them, is this the correct approach for this problem ? Thanks

How could I plot a graph of the random walk over each iteration?

Great video! Greetings from Argentina

@NormalizedNerd

3 жыл бұрын

Thanks mate! :D

Supose that we a huge A matrix (realy big dimension). Would be faster/cheaper to use the nontecarlo aproach rather than finding the eigenvectors?.

Thank you for your Video, I just want to know some more details of the last part. In how many iterations of the Markov chain do you expect that see the specific random walk? As we know by increasing the iteration of random walks it would be more chance to appear spesific random walk. can you please clear this part?

To quote South Park: "Pizza, french fries, pizza, french fries....see? he should of done pizza instead of french fries."

How do you determine the probability of the transition matrix?

U r the best

@NormalizedNerd

3 жыл бұрын

Keep supporting :D

can you make video on HMM practical implementation please

is it possible to trade NIFTY india using MARKOV chain ? mostly on the LONG side?

Method 2 curls my nails a bit. Instead of of multiplying A many times you could just square A again and again, which will lead to A^(2^n) after n steps. That would be so much faster.

@simonmasters3295

2 жыл бұрын

Good point, but at this level of matrix dimensionality it computes fast enough anyway...

can you make a full course on stats and coding? i would be wiling to pay for the service. thanks for the content.

can you make practical implementation video on HMM please

I don't know if you mentioned, but google's search algorithm uses these stationary distributions!!

@NormalizedNerd

3 жыл бұрын

Yeahh!...No, I didn't mention that in this video.

Is there a way of verifying the values of stationary state?

@simonmasters3295

2 жыл бұрын

He's just shown you three methods. If they agree, no problem. In an earlier vid if the system is oscilatory the stationary state will not settle.

How to run this

Hi. Awesome videos! Which college are you/were you from?

@NormalizedNerd

3 жыл бұрын

Kalyani Government Engineering College.

Hi. How to do it for time series data?

@simonmasters3295

2 жыл бұрын

Very good point, especially where the states have long lives or variation time between or in transition

I can has Baum-Welch algo next pls?

Why didn't you use FOR loops instead of WHILE loops?

@NormalizedNerd

3 жыл бұрын

Haha...no specific reason

Last time it was a hamburger and not a hot dog😂

@NormalizedNerd

3 жыл бұрын

Really? 😂😂