This is exactly what I was looking for since hours! This is the best video I found after a lot of hunting for simple intuition behind logistic regression.
Both videos 7 and 8 were excellent. I wish you had more like this that would help us understand from soup-to-nuts how the coefficients are built internally using real data in a simple-to-understand video that shows us how the betas are calculated.
Thank you so much, this is the best example driven teaching of this subject ever. I tried to understand this from books and multiple other videos, but didn't really get it it until I watched this video.
Find the logistic function f with the given properties. f(0) = 1, f has limiting value 16, and for small values of x, f is approximately exponential and grows by 75% with every increase of 1 in x.
Dear Professor I am a statistics student, search all Internet for a good explanation and found yours, that is the best video explanation I've ever see. So I have one doubt, in linear regression model we can evaluate the whole mode looking for a significant p-value, lower error and higher R2. How we can do it on Logistic Regression? I normally use R and see it produces only information below: Null deviance: 29.648 on 24 degrees of freedom Residual deviance: 25.198 on 22 degrees of freedom AIC: 31.198 How I can use them to evaluate if the whole model is good or not ? Thank you and congrats again for excellent explanation.
Question: Is it correct to say that the predictive power of Age is greater than Woman (just by looking at the absolute value of the coefficient)? If I were to remove 1 variable for any reasons, should I always remove Woman (assuming that the model is correct and the learned coefficients are optimal)?
Assuming that for example the coefficient " constant" is not significant..or any other coefficient... Would you then simply exclude it from the calculation of the probability or still included in the probability in your calculations?
Thank you very much for the videos you uploaded, well explained, one thing I am wondering is here we have two independent variables, age and sex, what its graph will look like comparing to the model with one independent variable age.
what if the ages are given in classes...like 15-24,25-34,35-44,...and you are asked to find the odds of a 25-34yr old living in a village, from your video i understand that i will multiply the other coefficients by one but what will the coefficient for the interested age class be multiplied with as the age is in a range and not a point estimate as you used in your example? should i subtract 25 from 34 and use the resultant difference to multiply its coefficient???
hi! may i ask, can the coefficient for woman in the multiple regression u have shown which is -0.557795 can be interpreted as "the likelihood that a woman subscribes in a magazine holding age constant is 55.78% less than a man"?
Great video! However, would be nice to explain also the intercept. Why is it negative? I know that the more negative the intercept is, the more the logistic curve is shifted to the right, therefore implying few y=1 observations..
What does that negative sign for woman coefficient meant....you told that women are less likely to subscribe to the magazine as sign was negative but their probability was increased from 35 yo to 36 yo...Kindly clear my doubt
What does training dataset contain? I mean, for \bf{x} I know, what is the training dataset's label? I mean, \bf{Y}? Because I want to know how do you get the linear model: y^* = \beta^T * \bf{x}. For the process from y^* to p I understand.
This is the best video explanation I've ever seen on this topic! Thank you sir!
so helpful, learned more in one video than one entire semester of 6 credit stats class.
This is exactly what I was looking for since hours! This is the best video I found after a lot of hunting for simple intuition behind logistic regression.
15:44 "woman turned on" Didn't expect a video on logistic regression to contain those very words, spoken matter-of-factly
Best Tutor I have ever met on youtube
I agree!!
Thanks, I needed a quick refresher on how to interpret LogReg coefficients. This is just what I was looking for. Well done!
Both videos 7 and 8 were excellent. I wish you had more like this that would help us understand from soup-to-nuts how the coefficients are built internally using real data in a simple-to-understand video that shows us how the betas are calculated.
This is amazing!! thank you so much, it's great help when writing a dissertation!! Saved me!!
Jesus H. Christ, i can understand, such good teaching technique, clean, crispy, useful, easy to understand
THANK YOU SO MUCH! This is the clearest explanation out there.
Watching this in 2023, best I've seen yes
Thank you so much, this is the best example driven teaching of this subject ever. I tried to understand this from books and multiple other videos, but didn't really get it it until I watched this video.
Thats what i was looking for! More clearly explained than in the Wooldridge book! Thanks a lot!
Thank you for a fantastic video on Logistic Regression !!
Great explanation of logistic regression. The video 7&8 make it much easier for me to understand. Thank you!
Thanks a lot for this amazing video. it explained really the concept of Logistic. THUMBS UP!!!
wonderful, very clear and straightforward
Lovely presentation! You are a wonderful teacher. Thank you so much for doing this for others who need help in Statistics!
Absolutely amazing. Thank you so much !
MUCH clearer. Thank you
This is a clear explanation. Thank you!
Excellent Teaching. Thank you sir.
nicely explained, easy to understand in-depth explanation
You are a great teacher
Your explanations are a Life Saver!! Thank you very much and keep up this excellent job!! :-)
thank you very much for your explain
Great and very helping video! Thank you!
Thank you so much. Very useful and awesome explanation!
Really helped me for my midterm. Thanks a lot!
best explanation out there, thank you kindly
Find the logistic function f with the given properties.
f(0) = 1, f has limiting value 16, and for small values of x, f is approximately exponential and grows by 75% with every increase of 1 in x.
Excellent Video. If you can create another one on how to estimate the coefficients and more examples on multiple logistic regression.
This is the best video explanation ..Thank you sir
Thanks a lot really very useful, very simple explaination and very insightful :)
Thanks so much, the video helping me to get more understand for the way the model comes from. Waiting, if you will produce more useful video. ^^
Awesome interpretation
Like using R for example, you can use lm() to mimic glm() logit.
Great! Thank you very much!
Wonderful ... Thanks a lot Sir!
Dear Professor
I am a statistics student, search all Internet for a good explanation and found yours, that is the best video explanation I've ever see.
So I have one doubt, in linear regression model we can evaluate the whole mode looking for a significant p-value, lower error and higher R2.
How we can do it on Logistic Regression? I normally use R and see it produces only information below:
Null deviance: 29.648 on 24 degrees of freedom
Residual deviance: 25.198 on 22 degrees of freedom
AIC: 31.198
How I can use them to evaluate if the whole model is good or not ?
Thank you and congrats again for excellent explanation.
Very helpful, thank you!
Great video! Even though it's in Excel, it helped me create a good solution in python to solve a related problem!
Question: Is it correct to say that the predictive power of Age is greater than Woman (just by looking at the absolute value of the coefficient)? If I were to remove 1 variable for any reasons, should I always remove Woman (assuming that the model is correct and the learned coefficients are optimal)?
Amazing! Thanks a lot
Helped alot! I suck at statistics and this video made everything clear. Have a mid term tomorrow.... :/
+Asif Irtiza Hussain mid valo hoisilo to vai? :v
+Avijit Roy haha... Tmio ek jagay eshe thekla? Bujchi... Report korteso
awesome!
Fantastic.. thanks alot
awesome explanation :)
How do you estimate the coefficients for logistic regression?
Great Clip...
thanks, sir.
Thank you sir...
Assuming that for example the coefficient " constant" is not significant..or any other coefficient...
Would you then simply exclude it from the calculation of the probability or still included in the probability in your calculations?
Thank you very much for the videos you uploaded, well explained, one thing I am wondering is here we have two independent variables, age and sex, what its graph will look like comparing to the model with one independent variable age.
Really helped alot sir..please share more videos
what if the ages are given in classes...like 15-24,25-34,35-44,...and you are asked to find the odds of a 25-34yr old living in a village, from your video i understand that i will multiply the other coefficients by one but what will the coefficient for the interested age class be multiplied with as the age is in a range and not a point estimate as you used in your example? should i subtract 25 from 34 and use the resultant difference to multiply its coefficient???
But if I just wanna the null hypothesis that considers only the intercept. No feature. What should I do ?
hi! may i ask, can the coefficient for woman in the multiple regression u have shown which is -0.557795 can be interpreted as "the likelihood that a woman subscribes in a magazine holding age constant is 55.78% less than a man"?
5 stars!!!!!
Great video! However, would be nice to explain also the intercept. Why is it negative? I know that the more negative the intercept is, the more the logistic curve is shifted to the right, therefore implying few y=1 observations..
What does that negative sign for woman coefficient meant....you told that women are less likely to subscribe to the magazine as sign was negative but their probability was increased from 35 yo to 36 yo...Kindly clear my doubt
What does training dataset contain? I mean, for \bf{x} I know, what is the training dataset's label? I mean, \bf{Y}? Because I want to know how do you get the linear model: y^* = \beta^T * \bf{x}. For the process from y^* to p I understand.
Do we really need to go for Logistic Regression for problems like character recognition ?
How can I find the name of this professor? I would like more content from him
is the reference category 1(=subscribe) or 0(=not) in this example?
Can you please explain Probit Regression model too?
how i get those at the beginning coefficients
I have a doubt. How did u lock in the values. to get ($C$3:$C$4)
+rithik marshall simply press F4 on your keyboard when selecting the range
Why are we multiplying the constant by 1?
That's because the assumed model is alpha*1+beta*var_1. alpha*1 is the constant term.
Amazing! May I know the name of the instructor and if he publishes any book/his own channel?
Plz give the link of data..
Do you really mean probability when you say you want to estimate the likelihood that a 35yo woman will buy the magazine?
How to get the values of alpha and beta plzzz anyone help me
I had started to think something was wrong with me...
wow
I feel offended. Can you do one with 72 genders? Ahah I am joking. Thanks a lot for the video :)