Hey danke für dein Feedback! Was hast du studiert? Wenn du magst kannst du dich ja mal per mail melden: mathias.jesussek@datatab.de 🙂 . Inzwischen trennen wir die deutschen von den englischen Videos, daher gibt es das gleiche sonst auch nochmal auf deutsch auf unseren deutschen Kanal : ) LG Hannah und Mathias
At 1:57, the false positive rate should be 2/5. If you are declaring diseased to be positive class, then showing healthy people as diseased is false positive. Am I correct?
I think there is a mistake at 5:36. It shouldn't say "the larger the AUC, the better the classifier," but instead, "the further the AUC is from 0.5." This is because 0 is not the worst classifier; 0.5 is (a random classifier). An AUC of 0 would actually be perfect since you could just invert the output of the classifier-meaning always pick "yes" if the classifier says "no," and vice versa. This would result in a perfect classifier.
Interesting point! From a purely logical point of view, you're definitely right, you could just invert the answer. But I think in diagnostic reality, an AUC of 0 would indicate that I have fundamental misconceptions (or accidental label swaps) in the model that need adjustment.
"True negative" means a healthy person is correctly identified as healthy (negative in this context means healthy). I think i's a bit confusing since we tend to associate "negative" with disease and "positive" with healthy, but it's the other way around.
Is it always true that the model with the larger AUC is the better classifier? Is there a case where a larger threshold will be used in application, and one ROC curve stays higher thru this threshold than a second ROC curve, even tho the first one may have a lower AUC than the second plot? Thanks!
Him thanks for that, and I have a question regards, the DATATAB, how to find the frequency, I have had tried multiple times, can't find it is there is ability to do it or find it in that? thanks
It would be "An" ROC curve because we pronounce the R as "ar". So that's a bit annoying since you say a ROC curve for the entire video xd. But this was a nice explanation thanks.
![Logistic Regression [Simply explained]](http://i.ytimg.com/vi/C5268D9t9Ak/mqdefault.jpg)
![Logistic Regression [Simply explained]](/img/tr.png)
![Hierarchical Cluster Analysis [Simply explained]](/img/n.gif)
If you like, please find our e-Book here: datatab.net/statistics-book 😎
It's been years since I tried to understand this concept, and finally with your video I get what ROC AUC is. sincere thanks.
حالا که این غلط گفت توی توضیحات ابتدایی. 😅
shouldn't false positive rate on 1:56 be 2/5 instead of 3/5?
I saw that mistake too!! you not alone!
So i am not alone, must be 3/5...
Best explanation I've ever had. Thank you.
❤
Words can't express my sincere gratitude, many thanks.❤
My pleasure 😊
Very very beautifully and simply explained Thanks
Most welcome 😊 Regards Hannah
no explanation can be better than this!
Thanks.
Fantastische Erklärung. Didaktisch ist das wirklich extrem gut. Respekt 👍🏻
Hey danke für dein Feedback! Was hast du studiert? Wenn du magst kannst du dich ja mal per mail melden: mathias.jesussek@datatab.de 🙂 . Inzwischen trennen wir die deutschen von den englischen Videos, daher gibt es das gleiche sonst auch nochmal auf deutsch auf unseren deutschen Kanal : ) LG Hannah und Mathias
@@datatab Hey. Ja sehr gerne. Ich habe Sozialwissenschaft mit dem Schwerpunkt Sozialforschung und Statistik studiert.
Best of all I have searched , Keep shining Friend 😀
you're also learning ML, are you not? haha....where are you from, friend?
At 1:57, the false positive rate should be 2/5. If you are declaring diseased to be positive class, then showing healthy people as diseased is false positive. Am I correct?
Oh, thanks for your comment! Yes you are correct! That's a mistake in the video! Thanks!
@@datatab so please pin this message.
Very well explained
This is a great explanation. Thank you
Beautiful explaination. Thank you !
I think there is a mistake at 5:36. It shouldn't say "the larger the AUC, the better the classifier," but instead, "the further the AUC is from 0.5." This is because 0 is not the worst classifier; 0.5 is (a random classifier). An AUC of 0 would actually be perfect since you could just invert the output of the classifier-meaning always pick "yes" if the classifier says "no," and vice versa. This would result in a perfect classifier.
Interesting point! From a purely logical point of view, you're definitely right, you could just invert the answer. But I think in diagnostic reality, an AUC of 0 would indicate that I have fundamental misconceptions (or accidental label swaps) in the model that need adjustment.
Great video
Glad you enjoyed it
perfect explanation.
Why is the false positive rate 3/5 and not 2/5 when 2 are wrongly classified as sick?
yeah, I agree with that, I think it should have been 2/5
great job , thank you
i think theres a mistake in this video at 2:44 where "true negatives" means diseased persons correctly classified as diseased
"True negative" means a healthy person is correctly identified as healthy (negative in this context means healthy). I think i's a bit confusing since we tend to associate "negative" with disease and "positive" with healthy, but it's the other way around.
I’m grateful
Is it always true that the model with the larger AUC is the better classifier? Is there a case where a larger threshold will be used in application, and one ROC curve stays higher thru this threshold than a second ROC curve, even tho the first one may have a lower AUC than the second plot?
Thanks!
Thank you for the video. I wondering how to get the 45 for the threshold value which is positive or negative
Easy explanation
Very Clear Explanation . Can you explain RoC for defaulter/ non defaulter ( altaman z score) and relate it to Type 1 error and type 2 error
Can you always construct a ROC curve for a logistic regression model?
Thank you so much
Him thanks for that, and I have a question regards, the DATATAB, how to find the frequency, I have had tried multiple times, can't find it is there is ability to do it or find it in that? thanks
It would be "An" ROC curve because we pronounce the R as "ar". So that's a bit annoying since you say a ROC curve for the entire video xd. But this was a nice explanation thanks.
False positive rate should be 2/5.. not 3/5 at 2.00 minutes of the video.. 3/5 is true negative rate
Also true positive rate should be 4/6, right?
you are right, she is misguiding us
Here comes the toppers 😂 seriously who cares man all you need is to understand the topic
truly amazing content in just 7 min video. hats off.
ie... that is!
The false positive rate will be 2 of 5
Good explanation ma'am, may i have your whatsapp no??
Many many thanks for your feedback! But unfortunately we do not give out our phone number!
@@datatab xD