bandit.Reward

bandit.Reward(self, outcome, covariates=None)

A simple class for reward functions.

Each reward function should return a reward object. For covariate adjusted algorithms, the reward should contain both the outcome as well as the corresponding covariates. For non-covariate-adjusted algorithms, only the outcome should be specified.

Attributes

Name	Type	Description
outcome	float	The outcome of the reward function.
covariates	pd.DataFrame | None	(Optional) The corresponding individual-level covariates.