Skip to content

Instantly share code, notes, and snippets.

@sergeyf
Last active February 3, 2018 22:51
Show Gist options
  • Save sergeyf/31c930fe9bbb5299282f to your computer and use it in GitHub Desktop.
Save sergeyf/31c930fe9bbb5299282f to your computer and use it in GitHub Desktop.
import numpy as np
class Data_generator(object):
def __init__(self,K,d,reward_type='binary'):
self.d = d # dimension of the feature vector
self.K = K # number of bandits
self.reward_type = reward_type
self.means = np.random.normal(size=self.K)
self.stds = 1 + 2*np.random.rand(self.K)
# generate the weight vectors
self.generate_weight_vectors()
def generate_weight_vectors(self,loc=0.0,scale=1.0):
self.W = np.random.normal(loc=loc,scale=scale,size=(self.K,self.d))
def generate_samples(self,n=1000):
# the X are only binary
X = np.random.randint(0,2,size=(n,self.d))
# the rewards are functions of the inner products with self.W
IP = np.dot(X,self.W.T)
# now get the rewards
if self.reward_type == 'binary':
R = (np.sign(np.random.normal(self.means + IP,self.stds)) + 1) / 2
elif self.reward_type == 'positive':
R = np.random.lognormal(self.means + IP,self.stds)
elif self.reward_type == 'mixed':
R = (np.sign(np.random.normal(self.means + IP,self.stds)) + 1) / 2
R *= np.random.lognormal(self.means + IP,self.stds)
return X,R
@lemondy
Copy link

lemondy commented May 12, 2016

hi sergeyf, I have read your three post whoes title is Personalization with Contextual Bandits. I am very curious about how you implement your simulation, can you open it? thanks.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment