Statistics 101: Logistic Regression, Odds Ratio for Any Interval
HTML-код
- Опубликовано: 22 авг 2024
- In this video, we learn how to calculate the odds ratio for any two values of the independent variable. We also graph the odds ratio change to fundamentally understand what is going on under the hood of Logistic Regression.
My playlist table of contents, Video Companion Guide PDF documents, and file downloads can be found on my website: www.bcfoltz.com
#statistics #regression #machinelearning
I have been consistently watching all 5 videos, hands down one of the best statistics tutorial playlist here on RUclips, can't believe this was posted almost 10 years ago!
This is one of the best video tutorials I've ever found on any statistics subject. I finally understand logistic regression!
Thank you, Brandon! I have watched all the Logistic Regression Series, and I feel that I am obliged to say thanks for all the effort you put into this series to make it elaborated like this!
You have done a great job to make anyone understands things from the bottom to the top!
This is what I call it "Job well done!".
Videos are good but I do have a suggestion, which is adding a number somewhere in the title to show the watch order of the videos and you could even show the total number of videos in the series eg 2 of 5. Initially when I found your videos I was confused as to where to start and I clicked on the third video where you referenced previous videos. I only managed to find out the proper order by watching them in the playlist you created however it would be more useful to have a number in the title cause other less resourceful or unmotivated people would just move on to other videos.
eliasm307 i agree.
@@Elrorosa agree :) !
THANKS is not enough.. i should say more than that, I started my day with logistic regression, planned to complete in 2 to 3 hours.. i had gone through lot of stuff which lead to confusion between logit and sigmoid function..... felt bad and sad, that i couldn't make it, then your stuff came, it is SAVIOUR.. You explained very clearly, detailed every thing in simpler manner. I ended my day with SMILE with your stuff.. APPRECIATE.. APPRECIATE.. APPRECIATE... Expecting stuff on Machine Learning Algorithms too.
I have watched all the videos related to logistic regression. I feel like a blind person feels if given the gift of sight. I was feeling totally lost. Thanks so much for doing such a great job of explaining something so complicated in such an easy and effective manner. You have saved me from failing!!! Well Done!!!
This is undoubtedly the best series on logistic regression I have found on RUclips. I understood everything as if my very life depended on it. I think the size of my brain must have increased by a few percentage points as well.
Bam! 18:52
I finally understand why I've been taught that exp(B1) = OR(B1)! Thanks!
Great learning videos.... I'm data mining PhD student.... this LR video is useful for my analysis... and this is very professional way to create videos.... Excellent service to humanity... keep it up Brandon
Thanks Brandon, I always come back to your videos for understanding the statistics concepts. You do have a knack of explaining in a way which makes it very easy and simple. Good luck !!
what a tutorial you got! man #1 videos i have ever seen!!!!
Brandon, Great work! I've learned a lot from your video!
I been looking for a clear explanation forever. FINALLY
You are genius Sir.You made statistics very easy by way of your explanation.None of my teachers were able to understand me in a way you did.Now Logistic regression is very clear to me.Hats off.. I am Masters student in Data Analytics
Truly outstanding play list--excellent work, Brandon. I took a SAS logistic regression course years ago and needed a refresher on logistic regression concepts. Your description of the concepts is far clearer than the one in the SAS course notes (which I still have) or in any of the statistics textbooks I have. Using the simple FICO score example to illustrate what the logistic regression output means, and how the numbers in the output tie back to the logit equation, makes the concepts crystal clear. I took screen shots during the videos. I'm sure I will refer back to them in the future. Thank you!
man this is the best tutorial i ever watched for this topic! my appreciation. I subscribed and gave u thumb-up.
Brandon, YOU ARE THE BEST!
Hi Sir Brandon, you are so thorough in your teaching, thank you
Your videos are simply the best on the topics you are teaching.. great job.
Appreciate the effort you are doing to educate the community. Lot of hard work goes into such videos. Full credit to you.
Hi Brandon,
These videos are excellent!! I am a business student doing undergraduate econometrics with no background what so ever in statistics so your videos have been a huge help to me, many thanks! Any chance you are working on some to do with regression for time series data, potentially covering auto-correlation? I realize things like auto correlation might go a little beyond statistics 101 but I am absolutely dieing to hear an intuitive explanation (you are the master at this) for things like that and heteroscedasticity. Everything I have heard so far falls well short of the way you explain things. You are an excellent teacher!!
Far more better than udemy course🔥 :)
Brandon must be an amazing instructor in the classroom.
really, you made my every concepts clear..😄😄.. thank you so much....
actually i am an engineering students and currently studying machine learning and had no clue about statistics .. but your tutorials helped me a lot..
once again thank you so much..
Great series of videos. Super helpful!!!
very good videos, been watching multiple linear regression and logistic regression playlists for my MVA partial and it helped me a lot.
By Max, Italian statistic student
If you had been my professor when I was a university student, I would have loved statistics. Ever since I hated Statistics so much.
But then when I watched your videos, I felt like oh 'why these videos never exist before.'
By the way, I am here because I am studying machine learning.
Thanks a lot. 😊😊😊
Excellent explanation, Brandon
Thanks a lot !!! These videos are very helpful. And the way you represent is really great.
Dear Brandon Foltz, I am a great fan of your statistical tutorial and it helped me to understand the basics very well. I would request you post a video on details over the various fields of the logistic regression and their meaning. Thanks for your great effort for making such great tutorials. Thanks Again!
Finally understand logit after watching your videos!!! Thank you so much!
Fantastic video on basic Linear Regression, it makes clear to everyone with a college level math. Concise and precise. I would humbly suggest some more details on maximum likelihood and, in general, on the math behind the scene.
Awesome videos! Thank you for your time spent on these; very helpful!
Excellent video! Well thought out content... even a lay man can interpret logistic regression output after watching this video.
Thanks for this video. It really did clear things up for me.
From last video:
Odd Ratio produced by tool = 1.0147.
15:30
FICO Change = 50
Odd Ratio Change = 1.0146^50=2.078
Odd Ratio vs X curve:
y = e^(0.0146 * x)
Odd Ratio increase (or decrease) vs X delta
y incr = e^(0.0146 * x-delta)
you are better than my tutor, thank you man!!
Excellent videos, very methodical explanations.
Thank you very much for your detailed and helpful videos. I have thought that odd and probability were the same thing for almost 4 year until I saw your videos =)) no wonder why I had so poor result in stat exams.
You are an amazing teacher!
Thank you!
very nicely put up great work brandon
I liked how in the multiple linear regression it was made very clear if you can continue the use of the statistical test by looking at the p-value and r2. Now with logistic regression if my p-value is higher then 0,001 should i not continue?
Excellent.
Thanks a million for sharing this...
Thanks! Your videos are great!
I have a request. Could you please cover Model Monitoring in one of your upcoming videos.
Cheer!
Amazing videos and lecture. I would appreciate if you could share the dataset. This would help people who don't use Minitab to do their analysis and have a hands-on experience on their software of choice.
Great Videos, keep up the great work dude!!
Extremely good material.
Please do a series on Multinomial Logistic Regression. Where one of the independent variables is categorical
Your videos are great!!
Hi Brandon,
Thankyou so much for this great knowledge sharing!!
Can you please share the sampe data set.
Hey Brandon,
Thanks a lot for the knowledge transfer that I have received through you and your videos.
Wanted to ask whether you will be teaching Decision Trees or Random Forest models etc?
Thanks,
Tarun
thank you; you made a tremendous difference
Thank you Brandon
Brandon awesome videos....very good explanation...
Good stuff Brandon.
Thanks for the videos.
these videos are helpful.. thanks
this was amazing! thank you!
These videos are so awesome! Thank you!
What programs did you use to create the animations? Just powerpoint or something else?
Thanks for the video, Brandon! I've just watched it for a second time. Quick question, I see Odds being really critical in both the concept and the formular here. I'm not really clear on the importance of Odds Ratio, other than just the ratio between two Odds. Could you pls help me understand that a little better? Thanks.
Great video Brandon! In all these videos you take a numeric variable(FICO score) but what if the features is categorical? which might have multiple classes? Even if you do one-hot encoding you still have a features with very limited range(when considering categorical variable)
I liked your videos. I am trying to find out what the intercept term in logistic regression means. All I found on the internet is that it the log odds when all variables are 0. Do you have a different explanation? Also to calculate the odds ratio when there are multiple variables do you substitute random values for the other variables not of interest when computing the probability and odds for the variable of interest, or do you leave out the other variables and essentially do a calculation like you did in this video with only one predictor variable?
I am missing a key piece of the estimated regression equation at 2:38 in the video. What is the value of e to get the sum of 1.20321477 in the numerator? I appreciate your clarity in making logistic regression and probability more understandable.
Found it!
What is it??/
Hi Brandon, great videos. Can you please explain what a coefficient is? I can understand that we would choose variables that are strong based on a high chi-square and put their coefficients in our equation. But what are they and how are they being calculated? Thanks.
Thank you for these very helpful videos :-). Can/should the Wald test be used to test for significance when performing binary logistic regression with data which I am treating as a population? I have been advised that inferential statistical tests are not appropriate with populations, so I'm confused whether to pay attention to the Wald test - particularly as it does not appear to have a straightforward relationship with the size of the coefficients/log odds. Thank you in advance for any answers!
The way you explain is good. But, what information are we getting from odds and odds ratio
YOU THE BEST.....
What a great video. However, can anyone explain how he got the exponential regression line in the video: 16.21. Why he didn't add "-9.346" (constant/beta zero) in the equation?
Thank you..
Hi Brandon, thank you very much for making such great video. I've been watching your videos from the beginning about logistic regression. I have a question about the range to get approved or not approved that score 630 (0.88) will not be approved and 640 (1.01) will be approved (you show at 11:20) When I compare them to the raw data (your video: Logistic Regression, Estimating the Probability) the score 655 is 0; 692 is 0; and 699 is also 0, etc. which mean that they are not being approved. I thought that by building this model, we will be able to know the score limit of being approved. I appreciate all answers to my question. Keep up the good work!
This is a regression model, it does not perfectly fit the training set of data.
Hello Brandon,
Hope you are doing well!
I saw most of your videos and all of them are very valuable for me to create more strong foundation in Statistics. I am looking for some simple and good videos related to credit risk modeling and methods used to calculate PD, LGD and EAD models.
Request you to please let me know if you have some links of the other tutors who can give me strong foundation in credit risk modeling?
Your prompt reply to my message will allow me to learn more and more via RUclips.
Sincerely,
Chirag
Hello! I wonder if anybody could be kind enough to please help me identify the datasets? Thanks in advance. I have looked for it in Brandon's blog, but I could not identify it, if its already there.
Also, thank you very much Brandon for your classes. I am starting with statistics and really appreciate them. They are really helping me.
how constant is negative and how do we interpret negative constant?
at ~11:38, why cant we conclude that +10 FICO score will increase your likelihood of approval 9.5% by just using 0.3966(phat of score 610) / 0.3622 (phat of score 610)
what does e stand for in y=e? how did you get 1.32 increase if 0.014634*19=0.2780?
Is there a video that explains 'e'? the exponential? Trying to do the math at 18:48 I couldn't figure out what 'e' meant or what was the value base. I had to do a lot of google search to find the base number. Is there a video that explains the concept?
e it’s a Euler’s number, approx 2,71828 but just punch in e on your scientific calculator and it will do the trick. Hope this helps 😀
Are you sure that natural log of 3 is 1.098? I am getting 1.58 and this is log base 2. Please clarify. Thank you.
does odd ratio and odds imply same thing? i.e. likelihood of happening something.
Is Odds ratio the same as likelihood ratio?
U forgot to put numbers 1,2,3 in logistic regression, to know high one to see first :)
Hi, I need your assistance
I have followed every video up to this series very well. They've all been clear and not too fast. However, this series completely lost me. Way too much dense information way too fast without enough context as to what's actually occurring beyond something to do with change in credit scores. Hope I don't need any of this info any time soon!