I built an uplift and cumulative gain charts curve and the benchmark is always a random model which is plotted at (0,0), (100%, 100%) (i e in an uplift model target 10 % of customers will provide 10% of positive responses , target 20% of customers will provide 20% of positive responses and so on).Here is the behavior of a random model which select customers without any criteria for selection of customers .
I wanted to know why the random model is always in the form of y = x , how we know that a random model will always do the same
here is a photo of how random model works , it is the one in pink that cuts the curve in the middle
I would be happy to have responses if anyone has a clear idea so that I can understand the idea behind (sorry for my poor english )
1 post - 1 participant