Support StatQuest by buying my book The StatQuest Illustrated Guide to Machine Learning or a Study Guide or Merch!!! statquest.org/statquest-store/ NOTE: A lot of people ask why, at 20:37, when we show the full tree that has Age > 50 as the first split in the tree, why the next split is Dosage >= 29 instead of Dosage >= 14.5. The reason it is 29 is that 29 is the best threshold after first splitting people on Age > 50. 14.5, the original threshold, was only shown to be optimal on the full dataset, prior to splitting the people into small subsets.
John, I just bought your book. Not because I need it, but because you deserve it. God bless you, you are godsent. you explain ML 10 times better than my professor
A picture (or video) is worth a thousand words. After scratching my head from looking at written formulas of how regression trees are created, this instantly cleared things up. Thank you!
I have completed the whole series on Trees till XGBoost! These are real treasure in teaching ML and trust me I have read a lot of books, ESL, Gilbert Strang, Larry Wasserman to name a few. I truly appreciate what you are doing and I will join your patreon!
OMG, I don't have words to appreciate this lecture. This is an impressive lecture I found for the Decision Trees algorithm. Thank you dear for your effort. I'm pretty sure you would come to need a lot of work to prepare for this lecture.
Statquest! Statquest! Can you note that you are the best! I have always hated statistics But now I love it coz of your tricks standard deviation and standard error After you taught it's no brainer I know these lines are cheesy But somehow you have made life easy :)
I have almost never commented or liked any videos on youtube but you are the one who made me proud to like your videos. you have beaten up all those paid and much-marketed course on machine learning and statistics A very thank you josh Starmer and tbh those bams do stick into people's head no lie.
The best video ever i have seen.. Trust me I was searching for random forest as regression concept all over internet and couldn't find as best as this one.... ❤️❤️
Josh, you are a blessing. I am filling all my gaps that my uni education thus far failed to do so and in that regard thank you for filling it in with something concrete that hopefully will stand for long and might grow strong by time. Stay happy man and the songs they kinda grew upon you, they are catchy.
Omg I can't believe the teaching has been done by real human. It feels like Robot is teaching. I will recommend this channel to all my friends, Thanks you so much!!
God level of explaination. This guy is doing marvolous job. You are a treasure for us, your channel is a one stop shop for us. Hats off for your efforts.
Tks a lot for the SUMMARY at the end sir, very much helps to quickly connect all the details explained in overall video, especially when the concept is in-depth and results in long video. Only a person with full understanding of concept (i.e only MASTER like you) can create such a simple perfect summary. My kind request to have summary section for all your upcoming videos as well please..... Thanks in advance and Thanks for this awesome content as well.
I'm the type of person who want to know true meaning behind all concept and if the statquest was not here I don't how much time would I waste to find all these. I can't thank u enough sir
How can anybody give dislike for this series of videos? 1) Other teachers from paid courses or from another channel 2) Egocentric people- "That´s too simple, you are insulting the statistics" 3) Naturally born haters 4) The ones that accidentally press the thumbs down instead of thumbs up QUADRUPLE BAM!
Honestly josh , After completing Machine learning in Coursera and Udemy , never got as much satisfaction after watching your videos , Iam redoing entire Machine learning from your videos.
I usually don't comment youtube vídeos, but I gotta say: this was very helpful for my data science study. Thanks for sharing your knowledge in a way that everyone can understand. Keep the good work!
I just can't believe someone can teach me ML in this greater depth for free! you really are super awesome Josh! Thank you for all the hard work you put
Treasure for anybody , interested to learn statistics and statistical learning.... thanks a lot josh the effort you put behind the lucid explanation ...
Excellent video for understanding all Regression Tree details. Did not believe it when I read "Clearly Explained", but it turned out very clear. Thank you
I'm having an exam in less than two weeks, these videos are super helpful! I wish I found you sooner. I'm actually understanding something, I feel less hopeless now lol. A big thank you!
00:05 Regression trees are used when fitting a straight line to the data is not useful. 03:07 Regression trees can accommodate multiple predictors and make accurate predictions. 06:05 Regression trees can easily handle complicated data and provide accurate predictions 08:46 The tree predicts drug effectiveness at 38.8, but the observed effectiveness is 100%. 11:49 Splitting data into groups based on thresholds with the smallest sum of squared residuals 14:51 Regression trees predict drug effectiveness based on dosage 17:48 Regression trees are built by finding thresholds that minimize the sum of squared residuals for each predictor. 20:11 Regression trees are a type of decision tree where each leaf represents a numeric value. Crafted by Merlin AI.
Thank you for your great explanation! I have a question: In your example of "building a regression tree with one variable" (i.e. dosage), the tree had several layers of branches split by different thresholds of dosage. Then moving on to your example of "building a regression tree with multiple variables", only one threshold of dosage was used (
Regardless of how many variables we have in our data, when we try to decide how to split the data at a branch in the tree, we always check every single one, and every single relevant threshold, and select the one that results in the lowest SSR. So, at any branch in the tree we check dosage, sex and age with a bunch of different threshold and use the one that gives us the lowest SSR.
@@statquest I think he meant "So, at any branch in the tree we check dosage, sex and age with a bunch of different threshold and use" MORE THAN ONE that gives us the lowest SSR, just like those you used in the example with single variable dosage.
I am currently trying to implement some regression ml methods to the data and these videos by Statquest are so simple and at the same time so informative, that if I don’t understand something, I come to find an explanation on Statquest. Thank you very much!
Question??? Q1) When we are using a bunch of predictors like dosage, age and sex. (At 22:32) Does predictors like age or dosage participate every time to become a node, even if they had been chosen earlier. For example (Age > 50) is already a root node, do we take age (as a predictor with different threshold) into account for branch node like (Age < 10)? a) Does it apply to classification (classification using decision trees) also where age contains numeric data, do we consider age every time when we find the lowest Gini impurity? b)And predictors like sex (categorical variables) don't participate more than once if they are the root node? At 20:00 Q2)What if we have a categorical variable with 3 outcomes, like instead of sex (male or female) we have hair color (Red Blue Green) now how we will calculate the threshold, will it be one vs all like (red vs others)?
A1) Yes. Features or Predictors can be used multiple time in the same tree. a) This is also true for classification trees b) Correct. A2) Usually it is one color vs. all others.
Thanks. Cant get better than this. Very likely that nowhere in the internet has this topic been conceptually explained so well and surprisingly in extremely simple terms. Because the other links are mostly just cryptic mathematical formulae, at best with a scratch on the underlying concepts.
Great content! I am reading ESL these days. And it can get really abstract and esoteric. Your videos make it much easier for me to understand the book. Thank you so much. And I bought your study guide too.
Please upload lectures on deep learning subjects especially computer vision. You have visualized all the concepts very effectively and deserves more than a typical university instructor.
BAM! Watched first video, subscribed, watched 4 more videos that were not relevant to my current quest ;) DOUBLE BAM! I think this is going to be my second home!
You are awesome man! I really like your videos and especially the way you make a complex concept very simple. The only thing I would like to see in the future is more videos on deep learning and machine learning.
Hola Perú. I'm glad my video was helpful. Good luck with your master degree! p.s. I have always wanted to go to Perú since I was a child and am trying to learn Spanish so I can go.
@@statquest Hey, Thanks for your wish, It will be an honor to show you my country if sometimes you come; my family and I will be pleased to host you if you want. My girlfriend and I speak english, perhaps we can be helpful.
How can these 10 people dislike this video. Kind of never satisfied soul. BTW, very well explained, thank you very much. I am becoming fan of your explanation... BAMMMMMM!!!!!!!!
Thank you very much for this video! It was fun to watch and I learnt a lot from the whole step-by-step process of building the simple regression tree :)
Question for the conditions: How the decision trees select the comparison operators to divide the data into groups. Suppose We have found out the best threshold then which operator (>, =,
Support StatQuest by buying my book The StatQuest Illustrated Guide to Machine Learning or a Study Guide or Merch!!! statquest.org/statquest-store/
NOTE: A lot of people ask why, at 20:37, when we show the full tree that has Age > 50 as the first split in the tree, why the next split is Dosage >= 29 instead of Dosage >= 14.5. The reason it is 29 is that 29 is the best threshold after first splitting people on Age > 50. 14.5, the original threshold, was only shown to be optimal on the full dataset, prior to splitting the people into small subsets.
Who came here after paying for ML courses on Udemy???? This explanation is far far better than those paid courses.
Thank you! :)
Me lol. That tutorial instructor is a real jackass
shhyitt you got me bro lololol.. since i can't just move on on some technical aspect without clear understanding
true story here
Absolutely u r correct
Screw Udemy, this guy explains better than my Machine Learning professor. You are the Bob Ross of statistics.
BAM! :)
whos is Rob Boss, the painter?
Couldn't agree more, Love you Josh
That’s the most beautiful I have read today.
This comment should be taken very seriously! Please don't spend money on Udemy - they sell crap! I am a victim of Udemy myself. Josh rocks!
Added more value than a fricking Harvard or Standford course in less than 15 minutes. Great job, I am forever greatful Josh !
Thanks!
You are just something else, I just can't believe how every time I return here, things that I used for years become suddenly clear.
Thank you very much. :)
Dude, you are a gifted teacher... well done, keep it up!
Thank you! :)
John, I just bought your book. Not because I need it, but because you deserve it. God bless you, you are godsent. you explain ML 10 times better than my professor
Wow, thank you!
A picture (or video) is worth a thousand words. After scratching my head from looking at written formulas of how regression trees are created, this instantly cleared things up. Thank you!
Hooray! :)
This is the best channel ever! I am a postdoc enjoying every single video ;)
Hooray! Thank you very much. :)
Same here!
Honestly, after every video I watch, I feel like I want to express gratitude. Clearly, you know how to explain clearly.
Wow, thank you!
I have completed the whole series on Trees till XGBoost! These are real treasure in teaching ML and trust me I have read a lot of books, ESL, Gilbert Strang, Larry Wasserman to name a few. I truly appreciate what you are doing and I will join your patreon!
Thank you very much!!!! I really appreciate your support and encouragement.
Are there any books recommended for an ML beginner? Thank you!
In Josh we trust!
What an awesome video Josh, you make ML understanding a pleasure...you are a gifted person and I am so happy you actually share.
Thank you very much! :)
Thank you for teaching important things like this for free. This is extremely good quality that is hard to find anywhere else.
Thank you!
OMG, I don't have words to appreciate this lecture. This is an impressive lecture I found for the Decision Trees algorithm. Thank you dear for your effort. I'm pretty sure you would come to need a lot of work to prepare for this lecture.
Thanks!
This has got to be one of the most entertaining educational videos ever made! You're just fabulous!
Thank you very much! :)
This was a better explanation by like 100 folds compared to explanation in my data science course for which I paid money.
I'm glad you like the video! :)
name your course please.
The search ends here. THere is no need to watch multiple videos to understand regression trees. this is the perfect video. Thank you JOsh!
Thank you very much! :)
Statquest! Statquest!
Can you note that you are the best!
I have always hated statistics
But now I love it coz of your tricks
standard deviation and standard error
After you taught it's no brainer
I know these lines are cheesy
But somehow you have made life easy :)
This is the best poem ever!!!!! Triple BAM!!!
Absolutely Amazingg
BAM!
@@statquest Hey cut back a bit guys.....I'm diabetic.
I have almost never commented or liked any videos on youtube but you are the one who made me proud to like your videos. you have beaten up all those paid and much-marketed course on machine learning and statistics A very thank you josh Starmer and tbh those bams do stick into people's head no lie.
Wow! Thank you very much! :)
Oh My God, Josh! Your channel is pure gold. I can not express how I like ur channel about all the explanations on complex ML concepts!
Thank you very much! :)
Yes same n I'm noob lol
man i know that decision trees are easiest machine learning concepts but you made them way easier than what it actually it is. Hats off
Thanks!
Immaculate teaching style. I love how you give a detailed intuition behind these otherwise complex topics!
Thank you very much! :)
The best video ever i have seen.. Trust me I was searching for random forest as regression concept all over internet and couldn't find as best as this one.... ❤️❤️
Thank you!
Josh, you are a blessing. I am filling all my gaps that my uni education thus far failed to do so and in that regard thank you for filling it in with something concrete that hopefully will stand for long and might grow strong by time. Stay happy man and the songs they kinda grew upon you, they are catchy.
Thank you very so much! :)
Omg I can't believe the teaching has been done by real human. It feels like Robot is teaching. I will recommend this channel to all my friends, Thanks you so much!!
Thanks for sharing!
I love how you sound like computer and make animation sounds too :D Your explanation is so much better than almost 90% of the rest .
Thank you! :)
God level of explaination. This guy is doing marvolous job. You are a treasure for us, your channel is a one stop shop for us. Hats off for your efforts.
Wow, thank you!
Tks a lot for the SUMMARY at the end sir, very much helps to quickly connect all the details explained in overall video, especially when the concept is in-depth and results in long video. Only a person with full understanding of concept (i.e only MASTER like you) can create such a simple perfect summary. My kind request to have summary section for all your upcoming videos as well please..... Thanks in advance and Thanks for this awesome content as well.
I'm glad you liked this summary!
Greate work. You might not know how good you are . Let me tell you. You are a gifted teacher.
Thank you very much!
Just found this channel recently, your explanation is better than any videos on youtube, it helps me a lot on my Data Science journey. Keep it up bro!
Thanks! :)
This is the best explanation of "Decision Tree regression" on youtube.
Thank you very much! :)
Today only I got to know "why decision trees". Your teaching is the best. blessings!
Thank you! :)
I'm the type of person who want to know true meaning behind all concept and if the statquest was not here I don't how much time would I waste to find all these. I can't thank u enough sir
Hooray!
How can anybody give dislike for this series of videos? 1) Other teachers from paid courses or from another channel 2) Egocentric people- "That´s too simple, you are insulting the statistics" 3) Naturally born haters 4) The ones that accidentally press the thumbs down instead of thumbs up QUADRUPLE BAM!
BAM! :)
Honestly josh , After completing Machine learning in Coursera and Udemy , never got as much satisfaction after watching your videos , Iam redoing entire Machine learning from your videos.
Wow, thanks!
I tried reading so many examples but your explanation is so good!! Keep teaching awesome concepts so that beginners like me can keep learning :D
Glad it was helpful!
I usually don't comment youtube vídeos, but I gotta say: this was very helpful for my data science study. Thanks for sharing your knowledge in a way that everyone can understand. Keep the good work!
Thank you very much! :)
why is he taking 14.5 as the root node???
@@shivamsehgal9120 Can you tell me what timepoint (minutes and seconds) in the video where you are getting confused?
I just can't believe someone can teach me ML in this greater depth for free!
you really are super awesome Josh!
Thank you for all the hard work you put
Thank you very much!
Treasure for anybody , interested to learn statistics and statistical learning.... thanks a lot josh the effort you put behind the lucid explanation ...
Glad you enjoyed it!
never stop making videos and saying BAM!
Thanks!
No one explained regression trees more clearer than him. Thank you 😊😊😊
Happy to help!
You've a talent for explaining things nice and clear. Wonderful. Thank you so very much!
Thank you! :)
This is the best channel for machine learning beginners. This help me so much, thank you alot! i will recommend your channel with my friends!
Thank you very much! :)
BAM!! Now I know Regression Trees without having to watch any other video.
Bam! :)
Excellent video for understanding all Regression Tree details. Did not believe it when I read "Clearly Explained", but it turned out very clear. Thank you
Thank you very much! :)
This is extremely well explained. I feel like I can explain clearly now too. Well done!!!
Glad it was helpful!
I'm having an exam in less than two weeks, these videos are super helpful! I wish I found you sooner. I'm actually understanding something, I feel less hopeless now lol. A big thank you!
Good luck with your exam! :)
Nice man, really nice. Unfortunately, I live in a hostile country and I could not pay for joining.
Just thinking about supporting me means that you are awesome! BAM! :)
Best video ever. I am going to become Patreon supporter. We need people like you to continue making videos.
Wow, thank you!
Hooray! I finally understand what is trees. Thanks for the explanation.
Hooray! I'm glad the videos are helpful. :)
00:05 Regression trees are used when fitting a straight line to the data is not useful.
03:07 Regression trees can accommodate multiple predictors and make accurate predictions.
06:05 Regression trees can easily handle complicated data and provide accurate predictions
08:46 The tree predicts drug effectiveness at 38.8, but the observed effectiveness is 100%.
11:49 Splitting data into groups based on thresholds with the smallest sum of squared residuals
14:51 Regression trees predict drug effectiveness based on dosage
17:48 Regression trees are built by finding thresholds that minimize the sum of squared residuals for each predictor.
20:11 Regression trees are a type of decision tree where each leaf represents a numeric value.
Crafted by Merlin AI.
:)
i feel luck that i found this channel .god bless you
Thank you! :)
Best channel ever for the stats and machine learning concepts!
Wow, thanks! :)
Thank you for your great explanation! I have a question: In your example of "building a regression tree with one variable" (i.e. dosage), the tree had several layers of branches split by different thresholds of dosage. Then moving on to your example of "building a regression tree with multiple variables", only one threshold of dosage was used (
Regardless of how many variables we have in our data, when we try to decide how to split the data at a branch in the tree, we always check every single one, and every single relevant threshold, and select the one that results in the lowest SSR. So, at any branch in the tree we check dosage, sex and age with a bunch of different threshold and use the one that gives us the lowest SSR.
@@statquest I think he meant "So, at any branch in the tree we check dosage, sex and age with a bunch of different threshold and use" MORE THAN ONE that gives us the lowest SSR, just like those you used in the example with single variable dosage.
The most clear explanation imaginable! Thank you, professor!
Thank you very much! :)
my lucky day because I was searching for regression tree and was wondering if statquest made a video on it already :)
YES! I'm glad you found my latest video. Support Vector Machines is next...
Wow…This is by far the best explanation about this topic. Thanks Josh
Thank you!
I feel like, why can't I cite this in my dissertation!!! Instead of citing those non-sensical papers and textbooks!!!
Bam!
This is the best channel I've seen in youtube!!
Thank you!
Using Sex to predict Drug effectiveness is missing one component: Rock and Roll! :-P
It's true, Rock and Roll is very important in this situation. ;)
He is the greatest teacher in the world.
Thank you!
I am currently trying to implement some regression ml methods to the data and these videos by Statquest are so simple and at the same time so informative, that if I don’t understand something, I come to find an explanation on Statquest. Thank you very much!
BAM! :)
Great video, love it, finally understand sum of square residuals and that it's not just a random number from a stat output. Keep up the good work!!
Glad it helped!
you have saved me so much time in life, i could cry
bam!
Question???
Q1) When we are using a bunch of predictors like dosage, age and sex. (At 22:32) Does predictors like age or dosage participate every time to become a node, even if they had been chosen earlier. For example (Age > 50) is already a root node, do we take age (as a predictor with different threshold) into account for branch node like (Age < 10)?
a) Does it apply to classification (classification using decision trees) also where age contains numeric data, do we consider age every time when we find the lowest Gini impurity?
b)And predictors like sex (categorical variables) don't participate more than once if they are the root node?
At 20:00
Q2)What if we have a categorical variable with 3 outcomes, like instead of sex (male or female) we have hair color (Red Blue Green)
now how we will calculate the threshold, will it be one vs all like (red vs others)?
A1) Yes. Features or Predictors can be used multiple time in the same tree.
a) This is also true for classification trees
b) Correct.
A2) Usually it is one color vs. all others.
@@statquest Thank you very much!!!
Thanks. Cant get better than this. Very likely that nowhere in the internet has this topic been conceptually explained so well and surprisingly in extremely simple terms. Because the other links are mostly just cryptic mathematical formulae, at best with a scratch on the underlying concepts.
Glad you enjoyed it!
BAM!!! I am addicted to these videos! Thank you so much, sir!
Bam! :)
Josh Starmer, man...
YOU ARE THE BEST FUCKING TEACHER ON THE PLANET! INFINITE BAM!!!!!!!!!!!!!!!!!!!
:)
Loved it ! Read several blogs, but no one can explain it like JOSH !!!
BAM! :)
Hey Josh your videos are truly amazingly couldn't wait for the video to end so I always hit the thumbs up before that.
Bam!
Thank you so much! Very well explained! And the pace is just right, not too slow and not too fast :)
Thank you! :)
Great content! I am reading ESL these days. And it can get really abstract and esoteric. Your videos make it much easier for me to understand the book. Thank you so much. And I bought your study guide too.
Thank you very much!! :)
Please upload lectures on deep learning subjects especially computer vision. You have visualized all the concepts very effectively and deserves more than a typical university instructor.
See: th-cam.com/video/CqOfi41LfDw/w-d-xo.html
BAM! Watched first video, subscribed, watched 4 more videos that were not relevant to my current quest ;) DOUBLE BAM! I think this is going to be my second home!
Hooray!!! Thank you very much :)
You are awesome man! I really like your videos and especially the way you make a complex concept very simple. The only thing I would like to see in the future is more videos on deep learning and machine learning.
Wonderful lecture! I liked the way you taught.
Many thanks!
Awesome explanation! Kudos!! I would repeat what others have said. You are an expert at simplifying complex concepts!!! Triple Bam!!!
Wow, thanks!
Amazing explanation Dr.Starmer, and the sounds you create while calculating values...they are indeed funny .. thanks
Bam! :)
Dude you made these topics bingeable! Great explainations!
Thank you!
Some of the best descriptions online...!
Thank you!
BAMMM!!!! from Perú, tnks a lot!! I´m a candidate of a Master degree and your videos make my day
Hola Perú. I'm glad my video was helpful. Good luck with your master degree! p.s. I have always wanted to go to Perú since I was a child and am trying to learn Spanish so I can go.
@@statquest Hey, Thanks for your wish, It will be an honor to show you my country if sometimes you come; my family and I will be pleased to host you if you want. My girlfriend and I speak english, perhaps we can be helpful.
@@HDelTrueno BAM! That would be awesome.
Thanks!
HOORAY!!! Thank you so much for supporting StatQuest!! BAM! :)
the best machine learning youtube video I have watched. Thank you statquest~
Thank you! :)
How can these 10 people dislike this video. Kind of never satisfied soul.
BTW, very well explained, thank you very much. I am becoming fan of your explanation... BAMMMMMM!!!!!!!!
Thank you very much!
Thank you very much for this video! It was fun to watch and I learnt a lot from the whole step-by-step process of building the simple regression tree :)
Glad it was helpful!
Thanks!
TRIPLE BAM!!! Thank you for supporting StatQuest!!! :)
Brilliant teacher. Impressed by the energy and passion
Thank you! 😃
You are an awesome teacher. Really appreciate the wonderful content. Also, the BAM’s never fail to make me laugh.
BAM! :)
Man, your videos are really helping me through my econometrics classes! Thanks!
Awesome! :)
BAAAMMMM! This is such a great video! I'm so glad I encountered Stat Quest!
Glad you enjoyed it!
God I love this channel so much, it makes complicated things which I couldn't understand during university so simple
Thanks!
You did a really great explanation. It's an easy way to learn ML algorithms. Thx a lot man!
Glad it helped!
Thank you for making these videos! After days of struggling ... Finally I am starting to get it 😭
Bam!
You have really relieved me. Thank you so much. God bless you.
Thank you! :)
Really excellent way of explaining the complicated concepts. Keep it up the great work!!
Thanks! :)
Simply brilliant explanation Josh! You make this so easy! TQ
Thank you! :)
12:59 This whole video in one sentence
BAM!!! :)
Clear and effective presentation. Great job!
Thank you!
Josh doesn't fail to amaze in each of his videos!!! Gosh JOSH for President :)
Hooray! And thank you for supporting StatQuest!!!!
Question for the conditions: How the decision trees select the comparison operators to divide the data into groups. Suppose We have found out the best threshold then which operator (>, =,
NOTE: Age < 50 and Age