def single_test(data_generation,
features,
actions,
outcome,
policy_factory,
matrices,
n_tests=1000):
print("Setting up simulator")
generator = data_generation.DataGenerator(matrices=matrices)
print("Setting up policy")
policy = policy_factory(generator.get_n_actions(),
generator.get_n_outcomes())
## Fit the policy on historical data first
print("Fitting historical data to the policy")
policy.fit_treatment_outcome(features, actions, outcome)
## Run an online test with a small number of actions
print("Running an online test")
result = test_policy(generator, policy, default_reward_function, n_tests)
print("Total reward:", result)
print("Final analysis of results")
policy.final_analysis()
header=None,
sep=" ").values
observations = features[:, :128]
labels = features[:, 128] + features[:, 129] * 2
import data_generation
import random_recommender
policy_factory = random_recommender.RandomRecommender
import reference_recommender
policy_factory = reference_recommender.HistoricalRecommender
## First test with the same number of treatments
print("---- Testing with only two treatments ----")
print("Setting up simulator")
generator = data_generation.DataGenerator(matrices="./generating_matrices.mat")
print("Setting up policy")
policy = policy_factory(generator.get_n_actions(), generator.get_n_outcomes())
## Fit the policy on historical data first
print("Fitting historical data to the policy")
policy.fit_treatment_outcome(features, actions, outcome)
## Run an online test with a small number of actions
print("Running an online test")
n_tests = 1000
result = test_policy(generator, policy, default_reward_function, n_tests)
print("Total reward:", result)
print("Final analysis of results")
policy.final_analysis()
## First test with the same number of treatments
print(
import data_generation as dg
import draw as d
from algorithm.km_smote import Over_Sample
from algorithm.under_sample import Under_Sample
if __name__ == "__main__":
dg = dg.DataGenerator(1000, 0.9)
data, label = dg.generate()
di = {"n_clusters": 10}
osampler = Over_Sample(data=data, label=label, n=3, categorical_features=[], **di)
syth = osampler.do_synthetic()
print("syth leng", len(syth))
syth_label = [1.0] * len(syth)
dr = d.Drawer(data, label)
dr.plot_scatter()
dr1 = d.Drawer(syth, syth_label)
dr1.plot_scatter()
under_sample = Under_Sample(major=data[0:900].tolist(), major_label=[0.]*900, synthetics=syth, synthetics_label=syth_label, categorical_features=[], rate=0.5, **di)
under = under_sample.do_undersample()
print("under length", len(under))
under_label = [0.0] * len(under)
x = under + syth
y = under_label + syth_label
dr2 = d.Drawer(x, y)
dr2.plot_scatter()
