Entropy (for data science) Clearly Explained!!!
Vložit
- čas přidán 19. 06. 2024
- Entropy is a fundamental concept in Data Science because it shows up all over the place - from Decision Trees, to similarity metrics, to state of the art dimension reduction algorithms. It's also surprisingly simple, but often poorly explained. Traditionally the equation is presented with the expectation that you memorize it without thoroughly understanding what it means and where it came from. This video takes a very different approach by showing you, step-by-step, where this simple equation comes from, making it easy to remember (and derive), understand and explain to your friends at parties.
For a complete index of all the StatQuest videos, check out:
statquest.org/video-index/
If you'd like to support StatQuest, please consider...
Buying my book, The StatQuest Illustrated Guide to Machine Learning:
PDF - statquest.gumroad.com/l/wvtmc
Paperback - www.amazon.com/dp/B09ZCKR4H6
Kindle eBook - www.amazon.com/dp/B09ZG79HXC
Patreon: / statquest
...or...
CZcams Membership: / @statquest
...a cool StatQuest t-shirt or sweatshirt:
shop.spreadshirt.com/statques...
...buying one or two of my songs (or go large and get a whole album!)
joshuastarmer.bandcamp.com/
...or just donating to StatQuest!
www.paypal.me/statquest
Lastly, if you want to keep up with me as I research and create new StatQuests, follow me on twitter:
/ joshuastarmer
0:00 Awesome song and introduction
1:28 Introduction to surprise
4:34 Equation for surprise
6:09 Calculating surprise for a series of events
9:35 Entropy defined for a coin
10:45 Entropy is the expected value of surprise
11:41 The entropy equation
13:01 Entropy in action!!!
#StatQuest #Entropy
Support StatQuest by buying my book The StatQuest Illustrated Guide to Machine Learning or a Study Guide or Merch!!! statquest.org/statquest-store/
Sir, have you included entropy in your book, I couldn't find it
@@yakubsadlilseyam5166 No, it's not in the book because, while it's nice to know about, it's not essential since there are other, easier to understand options that you can use.
@@statquest could you please provide sources to gain understanding of all types of entropies?
@@sandeepgill4282 en.wikipedia.org/wiki/Entropy_(information_theory)
It's a helpful introduction Josh....👍
This is quite possibly the best explanation of entropy I've ever seen. This is even better than Shannon's own paper!
BAM! :)
@@statquest Big Bang BAM
@@statquest DOUBLE BAM
What time Shannons was write that paper?
@@adipurnomo5683 I believe it was 1948
"Simplicity is the ultimate sophistication." - Leonardo da Vinci
Thanks for explaining in such simple terms!
My pleasure!
9 minutes of your video explained everything better than 2 hours of my professor giving a lecture... Thank you!!!
Glad it helped!
Make it 3 hours man
How'd the final go?
best entropy explanation. I took a 500 level ML class last spring in my masters and this is better explanation than my prof(no disrespect to him, he is amazing too)
Thanks!
You, sir, you are the very first person to actually explain this subject and not just repeat some random definition without giving any thought to it. I'm amazed by the amount of people who confuse rambling on about the topic with actually explaining it. Thank you!
Thanks!
Amazing video on Entropy Josh. Thank you. I am certainly more addicted to statquest than Netflix. I really liked the way you have introduced the notion of surprise, how you used it to pedagogically explain entropy. It is certainly now easy to think and remember the definition of entropy.
Awesome, thank you!!!!
@@statquest As a heavy CZcams addict with 40 sub-channels you are the only one I put a bell on
@@shahf13 BAM! :)
Absolutely amazing. I always come back to your videos from time to time for simple (but absolutely useful) explanation of complicated concepts that I found in papers. They all have helped me a lot, and I feel I'm better at communicating these concepts to other researchers too.
Hooray!!! I'm glad the videos are so helpful.
One of the best explanations of entropy. I had been struggling for a while with this concept and there was no intuitive way I could understand and remember the formula so far, your video made it possible. Great video!
Glad it helped!
Awesome as always, Josh! Thank you for continuing to share high quality content with us. You’re a very talented teacher. I wish you all the best!
Thank you very much! :)
I freaking love these video, Josh has a gift to explain things so vividly
Thank you very much! :)
Starting the video with a message declaring that we can understand Entropy is the best starting line I have ever seen from any teacher in my life. Great work!!!
Thank you very much! :)
😊😊15:45 hahaha “psst… the log of the inverse of the probability...”
Josh this was a fantastic tutorial. Love how I can just wake up and see content like this fresh in my YT recommendations
We all appreciate it
Thank you very much! :)
You can also say "negative log of the probability". Since 1/(x^c) = x^(-c), if c = 1, then log(1/x) = log(x^-1) = -1 log(x)
@@jeffnador9594 Yep. But, to me, that form makes it just a little bit harder to see what's going on.
@@statquest Agreed! But, the more you can keep people guessing, the higher the surprise value of the statement...
@@jeffnador9594 the point is to get an intuition and derive the formula. Not to manipulate mathematic terms.
Awesome world-level teaching...
Thank you!
Thanks!
Wow im sooo impressed. Frrr! It’s been a week trying to understand entropy and I rly thought I was never going to understand this bs. This video is arguably the best video on CZcams! Not only about entropy, but absolutely!! Thank you soo much
Glad I could help!
Your self promotion is not shameless, it a gift to humanity. Free content that explains way better than paid content on Coursera. Thanks for helping out poor guys like us Josh
Thank you very much! :)
so true
I'm truly amazed by the power of simplicity and intuition. Hats off Josh!
Thank you!
Loved it. The best explanation I came across on Entropy. Thanks Josh!
Glad you enjoyed it!
Thanks for all your efforts in creating a smart, funny and most importantly CLEAR explanation. This was awesome.
Glad you liked it!
I was awestruck when I finally understood what on earth Entropy is. In most algorithms, I hear entropy must be less, and I felt that it is some weird value which the model gives, and we have to tune it to reduce it. But now, sitting here, watching this video felt like an eye-opener. What a simplistic and beautiful way to explain complicated concepts. You truly are amazing, Josh!!! Super BAMMM!!
Thank you!
I follow most of your videos, not sure how I missed this gold! This is just the best entropy video I’ve ever seen.
Wow, thanks!
How can it be!! How can you simplify such complex topics into such simple explanations. Hats Off Man. I seriously wish if I could have had a Maths teacher like you back in school. I have become fan of your videos. Your videos are the first and last stop solution for all my doubts. Thanks Josh. You are a boon to learners like us. Impressed.
Thank you! :)
the best explanation for entropy i saw, after browsing for tens of videos. well done prof!!
Wow, thanks!
I love your videos, you make a wonderful effort for explaining difficult topics on a easy way. Huge thanks!
Thank you! :)
Thanks for these super clear explanations - you really manage to break down complex concepts till they seem simple and (almost) intuitive. Also really appreciate the pace, ie the patience of going step-by-step and not making any crazy leaps 2 mins to the end 👏👏👏
Thank you! :)
Please give Josh a nobel prize for not getting a single dislikes in many of his videos and for his contribution to the ML , Stats community. I have to forgive myself for not finding this channel way before when i started my interest in data science.
You are definitely inspiring me to teach many people the same content you taught us.
Thank you!!! However, this video actually has 13 dislikes. For some reason CZcams no longer posts the number of dislikes. However, with 3, 402 likes, that means 99.6% of the people like this video, which is pretty good.
@@statquest DOUBLE BAM!!!!
@@statquest The entropy, or the expected Surprise when you randomly pick a Like in your react pool, is pretty low :)))
@@buihung3704 bam!
czcams.com/video/ULc6z3EsXUs/video.htmlsi=wtYPowFO-N_iDqdS bazinga
This is THEE single BEST explanation for Entropy that I have ever heard. After this video I bought your book instantly. TRIPLE BAM!!
Wow! Thank you very much for your support!
this is definitely the most intuitive way to really gasp the ideal of entropy. you're just wonderful. Thank you soo much
Thank you very much! :)
I feel so happy because I just learned something new thanks as always!! 😁
I'm so glad!
me too!!!
Thanks, I’d been waiting for good explanation about entropy for a long time. Thanks :)
Thank you!
Absolutely amazing and intuitive explanation Josh! I couldn't have understood it without you, huge thanks :D
Thank you!
Been looking forward to this one for a while and was not disappointed at all - thanks!
bam!
You are the BEST teacher !! Thanking God that you exists.
Wow, thank you!
This is such a fascinating video, learning the theory of ML and I can certainly say you are a gifted person. Your perfect understanding field of Probability, Math and ML gives ability to explain it in the best way in the entire world. I'm amazed with your explanation skills
Wow, thank you!
"This is by far the best explanation for entropy that I have ever come across", thanks so much!
bam!
Really Great video. Right to the point. I met with one of my coworker who is very seasoned in Data science to help me work on a project and use entropy for the first time. After one hour I was as confused as I could be. But this video really helped. I wish if I saw this video before I had my meeting.
Glad it was helpful!
The inverse probability is a much better way of putting it than the minus sign. It was all this time in front of me and I didn't notice. Thank you!
Thanks!
Wow this was amazing. I've never seen entropy explained this clearly before
Glad to hear it!
Thank you so much for sharing this great explanation with us! I stopped the video after you derived the formula and then derived it again on my own. Makes so complete sense!!!
I am giving a talk at a conference in 2 weeks, and in my presentation there is a formula using mutual information. I was asked to explain this in my pratice presentation and was not able to. Now, after seeing your video, I am so clear about the concept of entropy and feel much more confident when I need to explain it :)
Good luck with your presentation! BAM! :)
This is a great explanation! Thank you Josh and StatQuest!
Glad it was helpful!
Hello, Thanks for this video, its really helpfull as always :D.
Quick question : why not use Surprise = 1- P(x), as it scale inverse to the probability and the surprise of a sure event is 0 as requested ?
Maybe it doesn't make sense for Surprise to be 1 when the probability is 0.
@@statquest I would expect a deeper answer than that, as the [0,1] range can be scaled with any constant to give any large-enough
surprise value to probability 0. It's just that the function would be linear now.
@@AAA-tc1uh Well, then you're stuck with the figuring out what that constant should be. Infinity? But that kind of opens another can of worms because anything times infinity is infinity. Thus, another advantage of using the log function is that the limit as x goes go zero is -infinity.
@@statquest Sure, I understand, but my rebuttal would be: we already use the [0,1] range for the probability distribution with 0="would never happen" and 1="always happens" (not entirely correct, I know, e.g. continuous distributions) so the same way we could treat Surprise value of 0 as "no surprise at all", and 1 as "maximum surprise". And we have a nice, well-behaved range with no infinities or undefined behavior.
Skimming Shannon's original paper I see he argues for the use of the logarithmic function in the opening paragraphs but never provides real deep reason other than convenience and and practicality in engineering usage (another point for the linear function suggested above). Edit: the real reason is the characterization of such function, see en.wikipedia.org/wiki/Entropy_(information_theory)#Characterization, which is apparently only satisfied by entropy function in this form, using logarithms, as proved by Shannon.
@@AAA-tc1uh Nice!
Fuck, I never understand about the entropy before watching this video, you are amazing
bam!
@@statquest double bam!!!
I have studied ml for a lot of time. i dont clearly know what is cross-entropy or entropy until now. This is the best explanation for entropy!
Thank you! :)
WOW its SUPEERRR EASY and well explained!! I only had known about the entropy in physics, but now I can see the calculation way of the entropy. THANK YOU for your hard work for easy understanding from S.Korea!
Hooray! I"m glad the video was helpful! :)
Being able to derive entropy and show it simply to us in a funny way is just mindblowing... Hats off to you sir!!
Thanks!
Hi Josh, you're always my go to for stats videos. Thank you, they're amazing. Would love to hear you talk through chi squared? I couldn't see anything on your channel thus far on it.
I'd like to cover that one day.
Best explanation i have ever seen. Thanks Josh. I don't forget this anymore. Keep teaching the people ❤
Simply amazing... I have been using CrossEntropy for months and only now I understood where the word Entropy came from in CrossEntropy
bam! :)
Just wow !
i've watched my prof's lecture (1.5 h long) 3 times and did not understand anything
and here you come with 15 minutes long video and BAM and medium BAM !! and I finally got it
THANK YOU!!!
BAM!
Hello! your videos are amazing and you do a great job in simplifying such difficult concepts and making them easier to understand. Thank you so much.
bam! :)
The world needs an army of people like you man. This explanation is outstanding. A triple bam.
Thank you!
I was NOT surprised by the high quality of this video given StatQuest's high probability of producing awesome videos.
Bam!
absolutely clear. Can't expect a more clearer explanation than this
Glad you think so!
I'm so happy these channel exists! hurray!!!
Thank you! :)
You are really great man!
You explained this so easily. Loved it!
Thank you!
DOUBLE BAM! Thanks for the video.
I liked how you derived the formula of entropy, and now it is easier to understand.
Hooray!
Thank you, Josh, you literally are the best teacher out there. I got a job as a Data Analyst and I Only watched your videos for all the explanations and understanding. Made a lot of notes as well. Sincerely Thank you.
PS: First thing I'm gonna do with my salary is buy a membership!! Infinite Bam!!
Congratulations!!!! TRIPLE BAM!!! :)
The best explanation of entropy! Thank you so much as always :)
Thank you!
Josh, thank you so much for your videos, you are a GREAT teacher. I wish you all the best, and thank you again!
Thank you very much! :)
You are truly amazing in simply explain the complicated things!
Thank you! :)
This is the best ever explanation of entropy I have seen!! The real surprise is the totally innovative idea of 'surprise! Thanks with entropy of zero!!!.
bam! :)
By far the best explanation I've seen. You are a true saviour.
Thank you!
Outstanding explanation made of simplicity. Huge thanks!
Thank you very much! :)
This is phenomenal work for ENTIRE DATA SCIENCE! Thank you a loot.
You're very welcome!
Bam!!! Hats off man that was the easiest explaination i ever come across...
Thanks for making such kinds of tutorials.
Glad you liked it!
This is absolutely amazing, I studied my whole course of Probability and statistics at my University but there was so much chaos. Now As I am learning machine learning this makes me a lot clear.
Glad it helped!
You might like to know that I am sharing this video with my whole class of CSC311 Introduction to Machine Learning at the University of Toronto.
You are doing a phenomenal work in explaining concepts in such an intuitively understandable way! Hugest thanks!
Thank you very much! I'm so happy the video is helpful! :)
It is very well explained about Entropy with simple example. Thank you so much. And most important you are not doing shamelessly promotion its actually worth it for everyone like me to understand the concept in very easiest manner.
Thank you! :)
I can’t thank you enough for your efforts and the way of teaching
You are very welcome!
Awesome awesome video. The best explaination of entropy I have seen so far!
Wow, thanks!
this is so smooth and easy to understand in connecting between “surprise” - an emotion with a mathmatic theory - number.
:)
This is a brilliant explanation! It all makes sense now!
Thank you very much! :)
YAY I have been waiting for a good explanation of entropy for a long time
BAM! :)
I bow down to you sir. It was truly amazing in a simple way.
Thank you!
Solid video, solid explanation. Best channel out there for your data mining needs
Thank you!
Hello from Brazil, thank you very much for this (the best so far) explanation.... I really impressed how simply you do the hole thing. Keep going.... Good Bless you.
Thank you very much!
Priceless channel for anyone interested in statistics. Just BRILLIANT.
Thank you!
I can't believe this guy made entropy fun and understandable. The intro song really lowered my anxiety about passing my exam thank you so much for your content
Hooray! :)
Sweet and perfect explanation, this was so insightful. thank you.
Thank you! :)
This is the BEST version of the explanation about entropy.
Thank you!
This is by far the best explanation I heard on entropy.
Thank you!
This man deserves a prize for how well he taught this. BAMMMM!!!!
Thanks!
You are awesome. I finally understood the entropy concept without any equations.
Big love😍
Thanks!
This is the most excellent explaination that I have ever seen. Impressive! Impressive! Impressive! Three times I must give. Thanks a lot!
Wow, thanks!
No surprise here: yet another amazing video! Great explanation!
Hooray!! Thank you!
wow! your explanation saved my butt big time! probably the best one on this platform
Thank you! :)
Fantastically clear explanation of a notoriously tricky subject. Apparently Johnny Von Neumann told Shannon to call his measure of information entropy, since “no one really knows what entropy is, so in a debate you will always have the advantage.” I suspect J V N wouldn't have said that if he's known about your video...
BAM!!! Thank you so much for your support!!! :)
I am surprised how easily understandable entropy is! Thanks!
Glad it was helpful!
This is the beeeeeeeeeest explanation of entropy! THANK YOU!!
Thanks!
It was just amazing .Thank you for that video!It really helped me.
I'm so glad!
Simple, elegant and funny. Just perfect. Thanks!
Thank you!
This is a purely amazing explanation. Bravo!
Many thanks!
Great video! For the first time, I understood what entropy means. Thanks!
Glad it was helpful!
This explanation is amazing. Thank you so much!!
Thanks!
The most complete video I've ever seen on this platform : ) I subscribed😎
Thank you!
Seriously, you are the best teacher I have ever had
Thank you!
Thank you for the great explanation. It’s very inspirational.
Thank you! :)
Simply Amazing. The best explanation out there 🔥🔥
Thank you! :)
Your videos are as straightforward as Y=mX. thanks Josh.
Glad you like them!
Excellent, thanks for the straightforward and extraordinary explanations.
Glad you enjoyed it!