MABpy package

Submodules

MABpy.ActionRewardAgents module

class MABpy.ActionRewardAgents.RandomAgent(verbose=0)[source]

Bases: MABpy.base.Agent

Choose random action and never learn

class MABpy.ActionRewardAgents.SimpleAgent(optimistic=0, verbose=0)[source]

Bases: MABpy.base.Agent

initEnviromentParams(params)[source]
class MABpy.ActionRewardAgents.UCB1Agent(optimistic=0, c=1.5, verbose=0)[source]

Bases: MABpy.ActionRewardAgents.SimpleAgent

class MABpy.ActionRewardAgents.UCB2Agent(optimistic=0, a=0.01, verbose=0)[source]

Bases: MABpy.ActionRewardAgents.SimpleAgent

initEnviromentParams(params)[source]
class MABpy.ActionRewardAgents.eGreedyAgent(greedy=0.1, optimistic=0, verbose=0)[source]

Bases: MABpy.ActionRewardAgents.SimpleAgent

class MABpy.ActionRewardAgents.enGreedyAgent(c=0.1, d=1.0, optimistic=0, verbose=0)[source]

Bases: MABpy.ActionRewardAgents.SimpleAgent

MABpy.GameEngine module

MABpy.MarkovianMAB module

MABpy.SimpleMAB module

class MABpy.SimpleMAB.BernoulliEnviroment(p)[source]

Bases: MABpy.base.GameEnviroment

class MABpy.SimpleMAB.DummyEnviroment(n_bandits)[source]

Bases: MABpy.base.GameEnviroment

class MABpy.SimpleMAB.GaussianEnviroment(n_bandits, min_mu=0, max_mu=1, min_sigma=1, max_sigma=1)[source]

Bases: MABpy.base.GameEnviroment

MABpy.base module

class MABpy.base.Agent(verbose=0)[source]

Bases: object

Base agent class. Agent makes decisions based on algorithm

Attributes:
_verbose - verbosity level _envParams - enviroment parametes
Learn(action, reward, context=None)[source]

Main learn function :param action: action :param reward: reward :param context: context vector :return: nothing

MakeDecision(context=None)[source]
initEnviromentParams(params)[source]

Save enviroment parameters :param params: enviroment params :return: nothing

class MABpy.base.EnvParams(n_bandits)[source]

Bases: dict

base enviroment

class MABpy.base.GameEnviroment(n_bandits)[source]

Bases: object

Base class for game enviroment

Attributes:
done - flag for end game params - public enviroment params
done = False
getBestAction()[source]
getBestAvgReward()[source]
getBestReward()[source]
getReward(action)[source]
getRewardSample(sample_size=100)[source]
params = None
reset()[source]
class MABpy.base.IteractionModel(verbose=0)[source]

Bases: object

Base class for agents-enviroments iteraction

Attributes:
_verbose - verbosity level
GetGameLogs()[source]

Get game logs :return: game logs

Play(max_iter)[source]

start iteraction :param max_iter: maximum iteration number :return:

Reset()[source]

Reset iteraction to init

Module contents