gistfile1.txt

"--cb_explore 2 --epsilon 0.2" 1 features 2 action 1 user even prob


Data format:

1:1:0.5 | feature1
2:0:0.5 | feature1

Loss for NO 2
E        ACTUAL: 0.11
E        DESIRED: -1.1

Prediction seem to be correct [0.9, 0.1]

What does it mean when predictions has n at the end
e.g. 0 n

"--cb_explore_adf --epsilon 0.2" 1 features 2 action 2 user even prob
Data format:

shared | User s_2
| feature1
2:0:0.5 | feature1
shared | User s_1
1:0:0.5 | feature1
| feature1

Loss for NO 4
E        ACTUAL: 0.48
E        DESIRED: -1.1

Prediction - actual values are [0.5, 0.5] is it correct?