StatQuest: Logistic Regression
HTML-код
- Опубликовано: 29 июн 2024
- Logistic regression is a traditional statistics technique that is also very popular as a machine learning tool. In this StatQuest, I go over the main ideas so that you can understand what it is and how it is used.
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 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...
RUclips 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
0:23 Review of linear regression
3:09 Logistic regression overview
6:22 Logistic regression vs Linear regression
8:06 Summary of concepts
#statquest #logistic
Support StatQuest by buying my book The StatQuest Illustrated Guide to Machine Learning or a Study Guide or Merch!!! statquest.org/statquest-store/
I've just bought the book. Thanks.
@@ivanjcardona Awesome! Thank you so much for supporting StatQuest! :)
Also bought the book and am very happy with it! I really recommend the book and love the idea of supporting the channel in such a way! Thanks for everything!
@@alexroode2659 Thank you very much!
Your videos helped me throughout my MS, helped me clear my interview for PhD at the 2nd best university of India (IIT Delhi), and now its helping me sail through my PhD. Words won't be enough for all the help you've provided! All I wanna say is Wish I can possess explanation skills like you and become a teacher like you someday!
Congratulations!!! Good luck with the remainder of your PhD.
Congratulations Subhadeep!
What's your branch ?
@@KumarHemjeet dept. of management studies
congrats on getting into IIT! My mentor is a grad of IIT-- what a university!
the intro killed me
:)
hahahahahah same
me too:)
bahahah same
@Phil Ad Ghost response
as far you're the best explainer I've ever seen.. Thank u Josh for all the efforts..
Thank you very much!!! :)
@@statquest xxx
Yesss
Outstanding! Its so rare to find people who can explain stats concepts in a simple, intuitive way. Well done!
Just want to say the quality of the video is top-notch. I'm currently pursuing MS in data science at a top-5 university in the US, but I did not find materials that explain logistic regression so clearly in such short space, and that managed to connect several related concepts together. Thank you for doing this!
Thank you very much! :)
Way too many materials pontificate.
"Consider a perpendicular set of data that is in concordance with the average to a set variance within a dataset that has parametric properties"
Translation:
"Here are residuals on data normally distributed".
It is actually quite shocking how many "advanced topics" when properly broken down and disseminated aren't nearly as difficult to learn.
Thank you so much for clarifying logistic regression in an easy to understand way! My prof is horrible at lectures, and this video really hits on the fundamentals of this topic
This is the best simplified explanation I have found, thank you! You gained a new sub!
Hooray! I'm glad you like the videos. :)
Mate, you are the only one who ties the knots all along the conceptual backgrounds which bring about the methods blindly used by most of us. Superb. Couldn't help but subscribe and comment here. Keep at it. This channel is pure gold.
Wow, thanks!
Thanks Josh! This is the most understandable material I've ever seen. Please keep up your excellent work! By the way, I had to replay the intro several times, love it!!
Thank you very much! :)
Videos like this make me wonder how these are free and those dry lectures at schools are charged
:)
Your videos are officially becoming my go-to when I want to learn stats. Thank you!
Hooray!!!
hats off to you to explain such a complicated topic in very easy sense (common sense) ...thums UP!!!
*BAM!*
*Thumbs up hahha Yes agree
Thank you so much for making statistics so interesting and EASY to understand!
You are really a very good teacher!
Thank you! :)
The most clearly explanation I ever heard!!!
Thanks! :)
I know it's an old video but I religiously watch your videos even today. I have become so used to it that even while taking notes, I write " BAM", "Triple BAM" and "Double BAM".
BAM! :)
My youtube watching history, you are the only guy who got like for every video I watch. and I like the videos which are genuinely helpful and amazing. You are the amazing stat guy every. Please, Keep doing what you are doing Josh!
Thank you very much! :)
This is so clearly explained, and helps clarify logistic regression from a conceptual point of view. thank you!!
Glad it was helpful!
just wanted to thank u again for all the efforts Josh :) ur Brilliant Lots of love & Respect !
Thank you so much!!! :)
just discovered your channel. It's amazing! Difficult consepts are explained so clearly
Thanks! :)
I SAW THIS IN MY LECTUREE AND I ALMOST CRIED THE SECOND I SAW STATQUEST UPLOADED YOU WONT UNDERSTAND HOW HAPPY I GOT YESSSSSS
BAM! :)
@@statquest DOUBLE BAMMM
OMG. This is the best channel for ML. Gonna recommend this to everybody in class. Hats off to your way of explaining such complicated topics in few min. Literally, I am more interested in learning ML in depth now all because of you. Why don't you drop a bootcamp course or something?
Thank you very much! :)
I'm liking the video just for the entry music. Relieved all my stress and I was able to focus better. What a positive way to start the lecture. Impressive man.
Thanks!
I would argue that this is one of the top 5 jingles of yours. I went back at least 10 times and listened to it. Double Bam!
Thank you Guruji!
Yes! I like like this one a lot too.
Thanks so much for all these brilliant videos, Josh! I've been binge-watching your vids lately and am learning so much from them. Speaking of fitting a squiggly line, any plans for ordered probit regression (which I believe too similar to logit regression?) and/or survival/hazard analysis?
Probit regression is on the to-do list. It is very similar to logistic regression - if you watch the "Details" videos that I made for Logistic Regression, I bet you can figure it out! ;)
What a brilliant explanation of a 'tricky' technique! Well done, Josh, and thank you so much - I'm helping my son with his MA in Pol. Sci.; I have taught (more elementary!) stats myself; I'd heard of LOGISTIC REGRESSION, but had never myself used or taught it; you've made it SOOO clear!!
Wow! Thank you very much! :)
This explanation is better than the thousands I've spent getting my MS. THANK YOU!
Glad it was helpful!
I didn´t understand Logistic Regression at all. I was desperate, then I clicked on this video...and BAAM. Love you. You really helped me. Thank u Josh!
Hooray! :)
That was the best explanation of MLE I ever encountered!!! I will use it in class! Thank you!!!
Awesome!!! :)
I totally agree, i finally understood MLE
WOw , The best expression. You rock ! thanks a lot
You’re welcome!!!
Josh, you are truly the best and most beautiful person on RUclips. Not just because you help me out of tricky situations, but just your general attitude makes me happy to learn statistics!
Hooray! Thank you so much! :)
you are the one of the best teacher that i am learning from, thank you stats quest!!!
Thank you! :)
Thank you so much for revealing in such a clear way the otherwise esoteric enterprise of statistics for non-statisticians / mathematicians! People like you makes all the difference in educating a lot of people.
Thanks!
Dead Josh!
I just randomly chosed this video from the many you uploaded to your channel to say that a few days ago I got my master degree from Survey Statistics and to say thank you for your awsome videos. You helped me a lot to absolve this challenge with sucess, thank you 1000times :)
Congratulations!!! That's awesome!! TRIPLE BAM!!!!
Thank you again for this video! This is my second time watching it (after a long while) and it is still so relevant and helpful!
Hooray! :)
Your videos are helping me out through my BAIM courses. Simplifying 20-pg chapters that are filled with mostly overly complex jargon helps so much
Happy to help! :)
Thank you so much for these tutorials and explanations! I love coming to your channel when I need to understand a machine-learning concept; I always find what I'm looking for. :)
Hooray! :)
You are helping me learn so much in such less time, that too without leaving any concepts. I prepared for my exam by watching your videos in 2 days for the course that was taught for 30 hours or so.
Good luck with your exam! :)
Best explanation on youtube! thanks and keep it up.
Thank you for your videos, they are so helpful!! I will be sharing them with my machine learning classmates!
Thank you very much! :)
omg, you're legit. thanks for the awesome explanation, instantly subscribed. helps that the intro is catchy as well
Thank you so much! :)
@@statquest OMG, you're "logit" :p
what an explanation!😊😁👏👏👌👌👌
You cleared all the ML doubts in my head... Thank you Josh
Awesome! :)
Best machine learning playlist out there ! Thanks for making such amazing content and helping students understand the concept. May god bless you ✨
Thank you!
The Intro Song is So nice 🙇🙇🙇
And actual concepts explained by you are on the 7th heaven
😎😎😎
Thank you very much! :)
I loved the intro so much I had to watch it a couple of times hahaha
Awesome!!! :)
You are amazing!! I just found your channel and I’m obsessed with it!
Thank you so much!!
Finally .. after reading alot about Logistic Regression make me even more confuse. i found this beautiful video.. easy to understand
Glad it was helpful!
Pressed the like button for the song first! I am sure the rest would be amazing as always! :)
Wow, thank you!
Done thanks
3:15 logistic regression predicts smth binary instead of continuous like linear regression
Logistic regression S shaped curve
Amazing work Josh. What a clear explanation. Vow. Great Work
Thank you very much!!! :)
thanks soooo much for the detailed explanation! i've been looking for it for a long long time and finally found it here!
Hooray! I'm glad the video was helpful. If you need more details, I've got videos that follow up on this one.
yes please! it would be great if we can see more videos about the complicated logistic regression model (ordinal, multinomial, etc) :)
Here are the videos that follow up on this one.
This video explains how the coefficients should be interpreted:
ruclips.net/video/vN5cNN2-HWE/видео.html
This is how the coefficients are determined:
ruclips.net/video/BfKanl1aSG0/видео.html
This is how you calculate R-squared and p-values for logistic regression:
ruclips.net/video/xxFYro8QuXA/видео.html
This next video is optional, but explains the concept of a saturated model and deviance statistics:
ruclips.net/video/9T0wlKdew6I/видео.html
Lastly I have a video on deviance residuals:
ruclips.net/video/JC56jS2gVUE/видео.html
The good news is that multinomial logistic regression is very similar to normal logistic regression. For example, if you have 3 possible outcomes (instead of 2, which is what you normally have), you simply look at the log ratio of the probability of the first option compared to the probability of the third option. Then you look at the log ratio of the probability of the second option compared to the probability of the third option. These ratios will tell you which option to to pick.
More good news is that I will be making videos to explain "regularization", which is a method used to solve for the maximum likelihood estimates for multinomial logistic regression.
thank you so much for those videos!!!! you have cured my fear of statistics!!!!
Thank you, very usefull explanation for ML newbies like me.
Hooray! :)
Another time gotta say that best teacher alive on the planet really the best teacher on the planet love you sir
Thank you so much!
Thank you for explaining the concepts so clearly!!
Thanks!
Anyone can make a simple topic difficult.. It takes skills to make a difficult topic simpler one.. You just made stats look easy to me.. That's the sign of a great teacher.. Keep up the good work..
Thanks a lot 😊
I am LMAO after the seeing the "Astrological sign" as a feature, awesome content Josh! I am learning so much from your videos and the content is helping me understand scientific papers!
Hooray! :)
I was looking for this comment :-)
Thank you for providing a short and fun explanation!!😁
Glad it was helpful!
very well explained, and super clear to understand, it is super clear, putting comparison together, to help us understand the differences, thank you
Thank you!
The introduction made me laugh hahaha you made my day thanks
This is one of my better intros. Thanks! :)
Ohhoo I am a 15 year old, and yet understood everything you said!!!
I now know Feed-forward neural networks and Convolutional ones
I'm focusing on LSTMs now
I've come so far because of your explanation on these basic algorithms (I mean Linear and logistic regression, and gradient descent)
Your explanations are the best!!
Keep it up!
Thank you!
Thank you very much! I'm glad my videos are helpful and wish you good luck with learning about LSTM Neural Networks. :)
Dare I say you make our research better and our minds clearer, thank you!
Thank you! :)
The thing i like most about this channel is that it reveals the inside details of the personal stat world, like in this video he revealed the expression "totes useless". Had no idea about this. Today I'm gonna use this in class and infront of my friends and everyone will know I'm the cool one.
Let me know how it goes! :)
This one video is already so much better than most of my online grad school classes.
bam!
you should definitely launch a channel just for your intro music. Quest On!
Why would ANYONE dislike your videos? Tells you how you can't please everybody no matter how good you are!
Thank you! :)
Thanks for your videos! I got super lost in class because my prof went straight to the math without explaining the concept of Logistic Regression first in English. You're a life saver.
Thanks!
I must admit one thing Josh. You've always made complicated statistical and ML concepts easier to understand. Thank you 🥰😊
Glad you think so!
That intro song deserves my subscription.
BAM! :)
The content is amazing in all videos; I loved this intro song the most than everything else :D
Thanks!
Damn that intro was good! i thought i clicked the wrong video, cause literally NO ONE sings for any educational video.. Mad respect!
bam!
The intros to these videos are slaying me! Great content though ;)
Ha!! I guess you have to take the good and the bad. ;)
Keep up the songs up man! They're slaying me in a good way, plus it take my mind off the Big Data exam pressure! :D
Will do! Good luck with the exam!! :)
Sir, this is for you 💛💛💛💛💛.....You helped me...I watched this video one night before the exam...and I nailed it.
Special mention for intro sir 💥💥
Thank you! I'm so happy you did well on your exam! :)
You explain so vividly.
Greetings from Nepal, SA. Thank you.
Thank you! :) p.s. When I was a kid I spent a summer in Kathmandu.
@@statquest Wow, glad to hear that sir, hope you visit Nepal sooner.
Keep on enlightening.
Do bless.
Nice! Simple explanation. Lots of graphs thanks!
Hooray! I'm glad you like it! :)
Those learning statistics need a hero. You are the one lol
Thank you! :)
Wheeeeeere the hell was this channel when I was studying :( smh. Brilliant stuff Mr. Josh Starmer
Thanks! :)
the best way of teaching. Thank you for this !!!!
Thank you! :)
You are the best explainer I've ever seen, thank you a lot !
Thanks!
Every time I watch your video, you remind me of Phoebe!! :P
Thank you for these amazing videos. Do you maybe plan to do more videos regarding generalised linear models like poisson regression or gamma regression? 😊 would be super cool!
I'll keep those topics in mind.
Hey Joshua.. well!! once again you nailed it. Thanks for giving such good insights. I was waiting for this. May be later if you can upload how to build models using some important Machine learning techniques. .!
Concise and easy to follow.
Thank you, Josh. You're the best!
Thanks!
Awesome explanation! Could you please make a video on multinomial regression as well. TIA!
I'll put it on the to-do list. :)
Thanks !! And I want to know how calculate the likelihood of logistic.
Awesome! :)
3 years late but I loved the intro, and the video too. Thanks a lot for putting in the effort :)
Thanks!
There people out in the world you just want to meet, like Josh! Great teaching sir.
Bam! :)
Hello,
I don't think the playlist for machine learning is in the correct order.
"astrological sign" haahhhahahahahahaha instantly subscribed
BAM! :)
I just wanted to have better understanding of ROC, picked a random youtube video... and now I feel like i need to watch every single video on this channel.
:)
Thank you for this video. You explained it super well.
Thank you!
logistic regression is predicting true/false as opposed to predicting something continuous like size ----> BAAM!
YES!!!! Hooray! Double BAM! :)
NO!!! It's not discrete. As said in video, its modeled by some sigmoid function, which has the property of giving output between 0 and 1 which is how probability is considered. You discretize it by saying values over some value of this sigmoid will be considered true/false. Triple BAM ! :(
I stand corrected! What 'n/a' said is true. Triple BAM! :)
Can't it be both though? Surely some outcomes are truly discrete - e.g. using alcohol consumption amounts to predict death. You could argue that being unconscious or in a coma are part of that scale but really you are dead or you aren't. Just like if a tumor is malignant or not, some tumors may spread faster than others but a tumor is either malignant or benign, not kind-of malignant but we'll just call it benign. I'm not an expert but that's how I thought of logistic regression (along with the classic examples of pass/fail a test which I agree is not really discrete, you've just arbitrarily set a cut off point on a continuous range of test scores).
@@sabrina76569 What you say is true - there are lots of outcomes that are truly discrete - and the last part of what you wrote is key - that you need some arbitrary cutoff to make an ultimate decision about dead/not dead sorts of things. Logistic Regression, along with all those other binary tests, first gives you a probability that one of those two things is true. Then you see if that probability is above/below a threshold. If above, then "dead", if below, "not dead".
I was watching this and my brother thought I was just watching memes lol
bam!
The most pricise and useful video i found on the topic
Thank you! :)
Thank you for such a concise and easy to understand video!
You're very welcome!
lmao the intro is the funniest gd thing ive ever heard. lol.
liked just for that intro
I'm in the 14th second of the video, and I'm already excited just coz of the informative intro soundtrack
BAM! :)
New sub! The explanation is straight to the point. Thanks a lot!
Thank you very much! :)