On the contrary, I find it super unclear. What exactly are the SCORE and COUNT? At 5:00 Score=0-100, Number of Positive=3 Number of Negative=62 What does it mean?!
@@karannchew2534 1-100 is the range of the score. 3 and 62 are positive and negative score that fall in that range.. But the cumulative calculation part is still unclear to me..can anyone please explain hoe to calculate that one..
It should not be accepted this 81.7% of Non-targets because cumulative count for score range of 400-500 for Non-target is given by (6212/7607)*100 = 81.7% (True Negatives) Plus, Amir is right, (100 - 81.7)% = 18.3% will be accepted as False Positives.
I believe Amir Shareghi Najar is correct. The cumulative distribution tells us that 81.7% of the non-target population scored in the ranges between 0-500. Therefore, if the threshold is above the 500 bracket, 1 - 81.7% = 18.3% should be accepted (The right side of the graph).
yes amir is correct. 18.3% of non target would be accepted. this can also be viewed as the false positive rate for this threshold value in the ROC curve
![Gunna - HIM ALL ALONG [Official Video]](http://i.ytimg.com/vi/w7Wf0h1Q8a4/mqdefault.jpg)
Thank you for the very clear examples. I was trying to refresh my understanding of the KS-test and it was super clear and well explained. Cheers!
On the contrary, I find it super unclear. What exactly are the SCORE and COUNT?
At 5:00 Score=0-100,
Number of Positive=3
Number of Negative=62
What does it mean?!
@@karannchew2534 1-100 is the range of the score. 3 and 62 are positive and negative score that fall in that range..
But the cumulative calculation part is still unclear to me..can anyone please explain hoe to calculate that one..
5:00 What exactly is the COUNT? The number of PREDICTED positive (targeted) and negative (non-targeted) outcome?
Very Nice Explanation Thanks :)
Shouldn't it be 100%-34,3% = 65,7% of targets will ve accepted ?
8:29 - 18.3% of non targeted would be accepted
It should not be accepted this 81.7% of Non-targets because cumulative count for score range of 400-500 for Non-target is given by (6212/7607)*100 = 81.7% (True Negatives)
Plus, Amir is right, (100 - 81.7)% = 18.3% will be accepted as False Positives.
I believe Amir Shareghi Najar is correct. The cumulative distribution tells us that 81.7% of the non-target population scored in the ranges between 0-500. Therefore, if the threshold is above the 500 bracket, 1 - 81.7% = 18.3% should be accepted (The right side of the graph).
yes amir is correct. 18.3% of non target would be accepted. this can also be viewed as the false positive rate for this threshold value in the ROC curve
Thank you all for the useful feedback!
thanks for your FB
i don't get how you calculate the cumulative counts. where do they come from?
how was the scoring done?
With a K-S equal to 47.4 is this no better than guessing?
Cumulative counts are wrong but the concept taught by you was good. Thanks #Aspiring data scientist