banditpylib.arms

Classes

• Arm: Arm

• PseudoArm: Pseudo arm

• StochasticArm: Stochastic arm

• BernoulliArm: Bernoulli arm

• GaussianArm: Gaussian arm

class banditpylib.arms.Arm(name: Optional[str])

Parameters

name (Optional[str]) – alias name

Inheritance

property name: str

Arm name

class banditpylib.arms.PseudoArm(name: Optional[str] = None)

Pseudo arm

This class is used to store empirical information of an arm.

Parameters

name (Optional[str]) – alias name

Inheritance

property em_mean: float

Empirical mean of rewards

property em_std: float

Empirical standard variance of rewards

property em_var: float

Empirical variance of rewards

reset()

Clear information

property total_pulls: int

Total number of pulls

property total_rewards: float

Total rewards obtained so far

update(rewards: numpy.ndarray)

Update information

Parameters

rewards – empirical rewards

class banditpylib.arms.StochasticArm(name: Optional[str])

Stochastic arm

Parameters

name (Optional[str]) – alias name

Inheritance

abstract property mean: float

Mean of rewards

abstract pull(pulls: Optional[int] = None) → Union[float, numpy.ndarray]

Pull the arm

When pulls is None, a float number will be returned. Otherwise, a numpy array will be returned.

Parameters

pulls – number of times to pull

Returns

stochastic rewards

class banditpylib.arms.BernoulliArm(mu: float, name: Optional[str] = None)

Bernoulli arm

Arm with rewards generated from a Bernoulli distribution.

Parameters

• mu (float) – mean of rewards

• name (Optional[str]) – alias name

Inheritance

property mean: float

Mean of rewards

pull(pulls: Optional[int] = None) → Union[float, numpy.ndarray]

Pull the arm

When pulls is None, a float number will be returned. Otherwise, a numpy array will be returned.

Parameters

pulls – number of times to pull

Returns

stochastic rewards

class banditpylib.arms.GaussianArm(mu: float, std: float, name: Optional[str] = None)

Gaussian arm

Arm with rewards generated from a Gaussian distribution.

Parameters

• mu (float) – mean of rewards

• std (float) – standard deviation of rewards

• name (Optional[str]) – alias name

Inheritance

property mean: float

Mean of rewards

pull(pulls: Optional[int] = None) → Union[float, numpy.ndarray]

Pull the arm

When pulls is None, a float number will be returned. Otherwise, a numpy array will be returned.

Parameters

pulls – number of times to pull

Returns

stochastic rewards

property std: float

Standard deviation of rewards