MLE vs OLS | Maximum likelihood vs least squares in linear regression
HTML-код
- Опубликовано: 4 окт 2024
- See all my videos at: www.tilestats.com
At 9:03 I should have said 4.24 and not 4.25.
1. Ordinary least squares (0:30)
2. Maximum likelihood estimation (03:41)
3. Log-likelihood (10:45)
4. MLE vs OLS (11:42)
This video was criminally low in my RUclips search results.
Мужик, спасибо. Лучее объяснение метода правдоподобия.
The BEST video explaining what MLE is and how we use Normal Distribution! Really thank you!!!
I'm gonna share the video a lot!!
the best about MLE in LR, even better than my teacher explanations
This video should be recognized more
Incredibly useful!!! Your first figure already explained more than my lecture slides.
Maybe the best explanation I came across so far
really appreciate your presentation format. it's always super clear and leaves little room for ambiguity.
Excellent explanations! Clear and easy to digest content.
Wow this is great. The visuals really help to reinforce what is the MLE equations. Thank you!
Bruh, it took me clicking on hundreds of videos to finally land on what I was looking for!
I hope you will consider doing a video on explaining what a link function is in GLM's and why they are used. I'm having trouble grasping the concept and how/why it relates to means and the normal distribution, etc.. I'm enjoying your videos very much, thanks again!
Have a look at around 7 min in this video about Poisson regression
ruclips.net/video/Obpz_Uvo2rQ/видео.html
@@tilestats Well, that was easy. Thank you! 😅
Wooow! It is the best lecture so far on this topic. Thanks greatly
Super! Excellent explanations! It is very nice and useful. Thanks.
Absolutely brilliant
This video is extremely good! You are great! you have new subscriber! keep going, well done!
Great vid. your lesson saves my life :D. Thanks for your dedication.
Fantastic video! Thank you :)
Thank you for the tutorial! 🙂
Thanks for the content ! FYI at 5:06 you probably wanted to point the arrow to the Y column, not the X one
True, thanks for spotting this error.
this is so awesome. ty!
Excellent
Thank you so much.
Do you have an opinion on Orthogonal Distance Regression ?
good explanation bro
4:48 Is this the visualization of: “residuals are normally distributed” for a linear model?
Yes
@@tilestats Thank you
Teşekkürler.
Thank you!
Thanks!
Thank you!
Hello. I sent you a question on your website few days ago asking about availability of the content of Multivariate statistics course in PDF format, and haven't received any reply from your side so far. could you please inform whether you plan to make them available on-line and when is the proposed date for this?
Thanks for your interest. I will try to do that in 2-3 weeks.
@@tilestats
Thank you.
It will really help as I have an exam of this course next month and your explanation is distinguished.
Hi
If you go to my homepage
www.tilestats.com
there should be a link under "Shop" to Payhip where you can buy the full course as 32 PDFs. Let me know how it works.
@@tilestats
This is perfect. Thank you a million times.
I have to inform you that I already bought 4 pdfs last week. Those that are related to Euclidean dustance and the PCA. Is there any chance to offset their price against the total price of the full Maltivariate course?
I can share with you the paypal e-mails regarding the transfer of the 4-lectures price.
@@tilestats
I have purchased it.
Thank you❤❤❤
Thanks a lot
Thanks for the video, TileStats. I just want to ask you if I understand the lecture video. First, to choose the right distribution, the distribution we choose such as normal distribution reflect the distribution of the target variable (i.e. data point).
Second, I am referencing to another lecture you created (ruclips.net/video/XpQp7gP6BnQ/видео.html). The distribution of prediction errors (i.e. error = the difference between the y and y^) has to appear to be normally distributed, if you choose the normal distribution. If you are using binomial distribution, the distribution of prediction errors has to appear to be binomially distributed. Can you please confirm this? Thank you!
When I refer to the distribution in this case, I mean the distribution around the line (the distribution of the residuals). However, in the special case of binomial distribution, where the response variable, for example, only can take two values (0 or 1), we use logistic regression where the model is fitted based on the ML method.
As an example where a normal distribution does not work, I suggest that you have a look at my video about Poisson regression and the videos about logistic regression:
ruclips.net/video/Obpz_Uvo2rQ/видео.html
ruclips.net/video/yhogDBEa0uQ/видео.html
ruclips.net/video/J0yuLu3oLuU/видео.html
Please do a GLM series, Also the subtitles on the bottom really help. Thank you!
GLM MLE Logistic and Poisson regression: ruclips.net/p/PLLTSM0eKjC2cYVUoex9WZpTEYyvw5buRc