• Search by Module
• Search by Words
• Search Projects

• Java
• Python
• JavaScript
• TypeScript
• C++
• Scala
• Blog

• striatum-master
  • .coveragerc
  • examples
    • movielens_bandit.py
    • movielens_preprocess.py
    • README.txt
  • LICENSE
  • simulation
    • simulation_linthompsamp.py
    • simulation_ucb1.py
    • simulation_linucb.py
    • simulation_exp4p.py
    • simulation_exp3.py
  • .editorconfig
  • .pylintrc
  • setup.py
  • circle.yml
  • setup.cfg
  • .travis.yml
  • README.md
  • requirements.txt
  • striatum
    • utils
      • misc.py
      • __init__.py
    • storage
      • recommendation.py
      • history.py
      • action.py
      • model.py
      • __init__.py
    • rewardplot.py
    • __init__.py
    • simulation.py
    • bandit
      • exp4p.py
      • linthompsamp.py
      • ucb1.py
      • __init__.py
      • bandit.py
      • tests
        • base_bandit_test.py
        • test_exp3.py
        • test_linthompsamp.py
        • test_linucb.py
        • __init__.py
        • test_exp4p.py
        • test_ucb1.py
      • exp3.py
      • linucb.py
  • .gitignore
  • docs
    • Makefile
    • make.bat
    • striatum.storage.rst
    • striatum.bandit.rst
    • striatum.rst
    • api_reference.rst
    • index.rst
    • conf.py
    • rtd-requirements.txt
  • tox.ini

"""Unit test for LinUCB
"""
import unittest

import numpy as np

from striatum.bandit import LinUCB
from striatum.storage import (
    MemoryHistoryStorage,
    MemoryModelStorage,
    MemoryActionStorage,
    Action,
)
from .base_bandit_test import BaseBanditTest, ChangeableActionSetBanditTest


class TestLinUCB(ChangeableActionSetBanditTest,
                 BaseBanditTest,
                 unittest.TestCase):
    # pylint: disable=protected-access

    def setUp(self):
        super(TestLinUCB, self).setUp()
        self.context_dimension = 2
        self.alpha = 1.
        self.policy = LinUCB(
            self.history_storage, self.model_storage,
            self.action_storage, context_dimension=self.context_dimension,
            alpha=self.alpha)
        self.policy_with_empty_action_storage = LinUCB(
            MemoryHistoryStorage(), MemoryModelStorage(), MemoryActionStorage(),
            context_dimension=self.context_dimension, alpha=self.alpha)

    def test_initialization(self):
        super(TestLinUCB, self).test_initialization()
        policy = self.policy
        self.assertEqual(self.context_dimension, policy.context_dimension)
        self.assertEqual(self.alpha, policy.alpha)

    def test_model_storage(self):
        model = self.policy._model_storage.get_model()
        self.assertEqual(len(model['b']), self.action_storage.count())
        self.assertEqual(len(model['b'][1]), self.context_dimension)
        self.assertEqual(len(model['A']), self.action_storage.count())
        self.assertEqual(model['A'][1].shape,
                         (self.context_dimension, self.context_dimension))

    def test_add_action(self):
        policy = self.policy
        context1 = {1: [1, 1], 2: [2, 2], 3: [3, 3]}
        history_id, _ = policy.get_action(context1, 2)
        new_actions = [Action() for i in range(2)]
        policy.add_action(new_actions)
        self.assertEqual(len(new_actions) + len(self.actions),
                         policy._action_storage.count())
        policy.reward(history_id, {3: 1})
        model = policy._model_storage.get_model()
        for action in new_actions:
            self.assertTrue((model['A'][action.id]
                             == np.identity(self.context_dimension)).all())

        context2 = {1: [1, 1], 2: [2, 2], 3: [3, 3], 4: [4, 4], 5: [5, 5]}
        history_id2, recommendations = policy.get_action(context2, 4)
        self.assertEqual(len(recommendations), 4)
        policy.reward(history_id2, {new_actions[0].id: 4, new_actions[1].id: 5})
        model = policy._model_storage.get_model()
        for action in new_actions:
            self.assertFalse((model['A'][action.id] == np.identity(2)).all())