I'm studying Data Science at MIT, you really can't imagine Aman how much "Unfold Data Science" is helping me, and a couple more channels, before I start any topic I like to tackle it first or just take a general idea, and you can't imagine how much your videos helped! Short, concise, and to the point! Thank you Aman 🙂
Please don't dislike, he is the bestest trainer, this shows that knowledge is power and maximum other RUclips videos are replica of one another making small modifications with no proper concept keep it up, you are the best
Excellent explanation. Esp the details on what happens when on feature is not selected and how it helps other features to vote in. Probably this also leads to feature importance too.
Dear Aman, thank you for your excellent explaination. As ai am a slow learner, I have a doubt from 11.25 mins. Is that the did advantages of Decision tree or Random Forest, because your video is the only source of my learning journey
Hi Aman . This is really great to see all the concepts in easy ,manner . Thanks for uploading it . I have a quick question , when we are testing our dataset on different decision trees then testing dataset will have all the N Columns and decision trees will have n1,n2,n3 columns then how it works ?
Hi Aman ...Till now your all videos are in order if following playlist from older to newer manner . Looks like now decision tree video should be part of this playlist after explaining ensemble and before random forest .... what do you think 🤔?
Hi Aman, While you say output of random forest is majority(suppose Y). Does that mean for all 300 inputs the prediction would be Y now. and for all test data the prediction would be Y only???
Its not like that you taking it wrong. Not all 300 data points will predict Yes. It shuffle data point row and column wise(not all 300 data points but 2/3 of the data). Its like if row no.1 given in bag1 and 2 other bags also with the corresponding to other feature. And bag1 giving output "Yes" and other 2 bags giving output "No" . Then it play a democratic rule which is every individual have same weightage and right to vote. So in this case the output will be "NO"
Hi Aman, thanks for your videos These are really informative and helpful I have one question I was asked by an interviewer When to use random forest instead of xgboost ?
Xgboost needs more server capacity on large data, random forest you get variablr importance, many more points to consider as well. This is a short answer
Good question Nidhi. There are two important parameters. One is "bootstrap" And other is "max_sample". For taking subset of data in each tree, you must say " Bootstrap " = True. By default it's True in python sklearn. Coming to "max_sample", if you say " none"(default), all records go in all trees. If you say a integer, those many rows. If you say a decimal, that percent of total no of rows.
I have one doubt, in which scenario i choose decision tree ML over random forest, because it seems random forest is the best , then why should we use Decision Tree classifier
sir, just a small doubt what are these decision trees in random forest classifier made of like do they have other classifier such as ann, logistic regression, svm and other types in them? is it so or something else
Ml generally is for predicting. Type of car, age of car, hours spent on the wheel, time of the day driver likes to work, economy of the city, and also drivers rating from other users
Sir for binary classification if no of tree are even number say we have 6 tree out of which 3 is yes or 1 and rest 3 is 0 or No Then what should b output of our Random forest method yes or No or else??
@10:22 You said that salary may not be part of further decision tree...(may not ) but what if salary is the only feature which has less entropy and high information gain. If it is so then i think in every decision tree root node will be salary only..... Or if it is taking different different rows and columns then i think it may happen that salary may not be always selected as a root node? i think i have question you also and answered my question by my own but you tell if im wrong then correct me please
Hi Shivansh, All the columns will not be selected in every tree. Hence, its possible that "Salary" is not part of few trees hence there is no question of it being root node.
@@UnfoldDataScience I should be the one to call you Sir lol. Regardless from where you are, your Data Science content is gold. The way you boil down and explain complex concepts in very simple English is really mind blowing. In shaa Allah planning to see all of your videos and extract maximum information from your channel.
I'm studying Data Science at MIT, you really can't imagine Aman how much "Unfold Data Science" is helping me, and a couple more channels, before I start any topic I like to tackle it first or just take a general idea, and you can't imagine how much your videos helped! Short, concise, and to the point! Thank you Aman 🙂
You just nail the big concepts with a simple example.
Thank you. Keep it UP.
Grow fast and furious!!
My pleasure Sagar.
wow what a teacher you are exceptional
i think no one on youtube can teach like you in so easy and lucid way
thank you sir
So nice of you Vishal.
Excellent explanation in simple English. Keep up the good work Aman! Thanks!
Welcome Harshal.
Beauty of this lecture is very easy and elegant explanation in simple English. deadly combination.. Thank you Aman
Welcome Prakhar.
#Interesting
Thanks a lot.
wow the example about salary in decission tree was sooo good! hats off
Your comment means a lot to me. Keep watching
Please don't dislike,
he is the bestest trainer,
this shows that knowledge is power
and maximum other RUclips videos are replica of one another making small modifications with no proper concept
keep it up, you are the best
Thanks a lot. Your words are precious for me.
Bravo! Excellent exposition.
Thanks a lot
I like your simplicity in teaching , you made topics simple. great job aman.
Excellent explanation in simple terms
Thanks Santhosh.
excellent content
Thanks Sangram.
Superb explanation , Keep going and growing. Thanks a lot.
Thanks again Sunil :)
Useful information....nice presentation
Thank you.
Excellent explanation. Esp the details on what happens when on feature is not selected and how it helps other features to vote in. Probably this also leads to feature importance too.
Thanks for watching
Well explained sir
Thanks Syed :)
wpw!!! what a gift in teaching!!!
This is pretty good sir. got a lot of input from this video
Great explanations .. thank you very Much.. Sir
Excellent presentation and content in a simplified way and shortest time ! Kudos to you. Thank you
Thanks Imran.
my great teacher, thanks
You are welcome!
Thank you
You're welcome
very great and clear
Glad it was helpful!
Dear Aman, thank you for your excellent explaination. As ai am a slow learner, I have a doubt from 11.25 mins. Is that the did advantages of Decision tree or Random Forest, because your video is the only source of my learning journey
Disadvantage of Decision tree - Overfitting
Disadvantage of Random Forest - Resource intensive algorithm
Great Explanations Aman.
Thanks Nikhil
Very good explanation ❤
Super lecture, easy to understand, keep up the good work bro...
Keep watching Srinivas.
You have explained the subject very well!!
Thanks a lot for the feedback.
finished watching
You are a great teacher!
Thanks a lot.
great video , simple and easy to understand , Thank you sir !
You are welcome. Keep watching:)
Hi Aman . This is really great to see all the concepts in easy ,manner . Thanks for uploading it . I have a quick question , when we are testing our dataset on different decision trees then testing dataset will have all the N Columns and decision trees will have n1,n2,n3 columns then how it works ?
Very good question - its not a parametric model so it does not matter.
Hi Aman ...Till now your all videos are in order if following playlist from older to newer manner . Looks like now decision tree video should be part of this playlist after explaining ensemble and before random forest .... what do you think 🤔?
Thanks for feedback Kirti. Let me check if I can rearrange. Happy learning. tc
Thank you Aman!
Welcome!
well explanation!!
Glad it was helpful!
Well explained, Thank you👍
Glad it was helpful Archana.
very helpful, keep up the good work !
Thanks Mehdi, will do! Stay Safe. Tc.
Good ..aman
Thank you.
Thank you :)
Hi Aman,
While you say output of random forest is majority(suppose Y). Does that mean for all 300 inputs the prediction would be Y now. and for all test data the prediction would be Y only???
Its not like that you taking it wrong. Not all 300 data points will predict Yes. It shuffle data point row and column wise(not all 300 data points but 2/3 of the data). Its like if row no.1 given in bag1 and 2 other bags also with the corresponding to other feature. And bag1 giving output "Yes" and other 2 bags giving output "No" . Then it play a democratic rule which is every individual have same weightage and right to vote.
So in this case the output will be "NO"
Hi Aman, thanks for your videos
These are really informative and helpful
I have one question I was asked by an interviewer
When to use random forest instead of xgboost ?
Xgboost needs more server capacity on large data, random forest you get variablr importance, many more points to consider as well. This is a short answer
Can you make videos on linear regression and logistics regression.
Hi Oyster, These videos are already there on my channel. Please find link below:
ruclips.net/video/8PFt4Jin7B0/видео.html
Dear sir, are there are two methods of constructing random forest algorithms ?
Sir , very nice video. Do you also take paid course?
Sir how can we decide which catagory is to be taken as the root node of any decision tree when more than 2 catagory is given in data
Could you rephrase your question please?
Thank u
Thank you.
Hello Aman can you please explain what it the difference b/w random forest classifier & extra tree classifier?
Hi Sager, for each feature , a random value is selected for the split in case of extra tree. I will explain in more detail in a video. Thanks you
I have a que. if we have 1000 of record data nd we build random forest and n_estimaters=10,then in each decision tree how many record will get train
Good question Nidhi.
There are two important parameters. One is "bootstrap" And other is "max_sample". For taking subset of data in each tree, you must say " Bootstrap " = True. By default it's True in python sklearn.
Coming to "max_sample", if you say " none"(default), all records go in all trees.
If you say a integer, those many rows.
If you say a decimal, that percent of total no of rows.
@@UnfoldDataScience tnx sir
for example if there are 500 decesion tress then it predicts 250 1 s and 250 0 s what the random forest will declares sir??
Highly unlikely scenario.
I have one doubt, in which scenario i choose decision tree ML over random forest, because it seems random forest is the best , then why should we use Decision Tree classifier
Normally we use random forest or boosting directly. No decision tree
if i have a more than 2 classes what to do
How to choose number of samples
sir, just a small doubt what are these decision trees in random forest classifier made of like do they have other classifier such as ann, logistic regression, svm and other types in them? is it so or something else
Decision tree.
Is random forest only for predicting? I’m tying to see which features affect the income of taxi drivers in NYC. Can I use random forest for that?
Ml generally is for predicting. Type of car, age of car, hours spent on the wheel, time of the day driver likes to work, economy of the city, and also drivers rating from other users
I don't think the sample has to have less observations. We sample N times for N rows of data
Both options are there
Sir for binary classification
if no of tree are even number say we have 6 tree out of which 3 is yes or 1 and rest 3 is 0 or No
Then what should b output of our Random forest method yes or No or else??
I think it just desides by tossing a coin🤔.
What is pasting?
There is no doubt that RF is much better than Decision Tree, then why still Decision Tree still in use ?
Not in use mostly.
Is there a proof of random forest’s accuracy as an algorithm? Thanks
Breiman 2001 has it
sklearn will give that.
@10:22 You said that salary may not be part of further decision tree...(may not ) but what if salary is the only feature which has less entropy and high information gain. If it is so then i think in every decision tree root node will be salary only.....
Or if it is taking different different rows and columns then i think it may happen that salary may not be always selected as a root node?
i think i have question you also and answered my question by my own but you tell if im wrong then correct me please
Hi Shivansh, All the columns will not be selected in every tree. Hence, its possible that "Salary" is not part of few trees hence there is no question of it being root node.
Ok u mean to say that in randome forest all the columns are not get selected at once for all the decision tree...columns gets selected randomly?
Yes absolutely.
are you from Rajistan?
No Sir.
@@UnfoldDataScience I should be the one to call you Sir lol. Regardless from where you are, your Data Science content is gold. The way you boil down and explain complex concepts in very simple English is really mind blowing. In shaa Allah planning to see all of your videos and extract maximum information from your channel.