From 36f544a88bb99e9a0a237d5a1ab5c105e5667ad8 Mon Sep 17 00:00:00 2001 From: usaito Date: Tue, 2 Feb 2021 07:15:32 +0900 Subject: [PATCH] update examples --- .../evaluate_off_policy_estimators.py | 2 - examples/obd/README.md | 2 + examples/quickstart/synthetic.ipynb | 47 ++++++++----------- 3 files changed, 21 insertions(+), 30 deletions(-) diff --git a/examples/multiclass/evaluate_off_policy_estimators.py b/examples/multiclass/evaluate_off_policy_estimators.py index bd0ca07b..8719be06 100644 --- a/examples/multiclass/evaluate_off_policy_estimators.py +++ b/examples/multiclass/evaluate_off_policy_estimators.py @@ -48,8 +48,6 @@ SelfNormalizedInverseProbabilityWeighting(), DoublyRobust(), SelfNormalizedDoublyRobust(), - SwitchInverseProbabilityWeighting(tau=1, estimator_name="switch-ipw (tau=1)"), - SwitchInverseProbabilityWeighting(tau=100, estimator_name="switch-ipw (tau=100)"), SwitchDoublyRobust(tau=1, estimator_name="switch-dr (tau=1)"), SwitchDoublyRobust(tau=100, estimator_name="switch-dr (tau=100)"), DoublyRobustWithShrinkage(lambda_=1, estimator_name="dr-os (lambda=1)"), diff --git a/examples/obd/README.md b/examples/obd/README.md index 34fdbd08..a192100d 100644 --- a/examples/obd/README.md +++ b/examples/obd/README.md @@ -5,6 +5,8 @@ Here, we use the open bandit dataset and pipeline to implement and evaluate OPE. Specifically, we evaluate the estimation performances of well-known off-policy estimators using the ground-truth policy value of an evaluation policy, which is calculable with our data using on-policy estimation. +Please clone [the obp repository](https://github.com/st-tech/zr-obp) and download [the small sized Open Bandit Dataset](https://github.com/st-tech/zr-obp/tree/master/obd) to run this example. + ## Evaluating Off-Policy Estimators We evaluate the estimation performances of off-policy estimators, including Direct Method (DM), Inverse Probability Weighting (IPW), and Doubly Robust (DR). diff --git a/examples/quickstart/synthetic.ipynb b/examples/quickstart/synthetic.ipynb index f202bc08..83075306 100644 --- a/examples/quickstart/synthetic.ipynb +++ b/examples/quickstart/synthetic.ipynb @@ -324,14 +324,14 @@ "output_type": "stream", "name": "stdout", "text": [ - " mean 95.0% CI (lower) 95.0% CI (upper)\nipw 0.787104 0.771925 0.807702\ndm 0.644029 0.642926 0.645100\ndr 0.779419 0.771589 0.788061 \n\n" + " mean 95.0% CI (lower) 95.0% CI (upper)\nipw 0.788252 0.770543 0.808750\ndm 0.643980 0.642763 0.645292\ndr 0.779467 0.769918 0.789674 \n\n" ] }, { "output_type": "display_data", "data": { "text/plain": "
", - "image/svg+xml": "\n\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", + "image/svg+xml": "\n\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", "image/png": "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\n" }, "metadata": {} @@ -365,14 +365,14 @@ "output_type": "stream", "name": "stdout", "text": [ - " mean 95.0% CI (lower) 95.0% CI (upper)\nipw 0.719964 0.705208 0.735488\ndm 0.627243 0.626011 0.628451\ndr 0.721119 0.713220 0.729274 \n\n" + " mean 95.0% CI (lower) 95.0% CI (upper)\nipw 0.719770 0.701483 0.738943\ndm 0.627292 0.626105 0.628534\ndr 0.722014 0.715064 0.731309 \n\n" ] }, { "output_type": "display_data", "data": { "text/plain": "
", - "image/svg+xml": "\n\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", + "image/svg+xml": "\n\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", "image/png": "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\n" }, "metadata": {} @@ -406,14 +406,14 @@ "output_type": "stream", "name": "stdout", "text": [ - " mean 95.0% CI (lower) 95.0% CI (upper)\nipw 0.606821 0.603119 0.610232\ndm 0.607460 0.605850 0.608856\ndr 0.607458 0.604074 0.610561 \n\n" + " mean 95.0% CI (lower) 95.0% CI (upper)\nipw 0.606659 0.603537 0.609438\ndm 0.607382 0.606054 0.608549\ndr 0.607823 0.605004 0.610869 \n\n" ] }, { "output_type": "display_data", "data": { "text/plain": "
", - "image/svg+xml": "\n\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", + "image/svg+xml": "\n\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n\n", "image/png": "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\n" }, "metadata": {} @@ -481,13 +481,13 @@ "source": [ "# we first calculate the policy values of the three evaluation policies using the expected rewards of the test data\n", "expected_rewards = bandit_feedback_test['expected_reward']\n", - "ground_truth_ipw_lr = np.average(expected_rewards, weights=action_dist_ipw_lr[:, :, 0], axis=1).mean()\n", - "ground_truth_ipw_rf = np.average(expected_rewards, weights=action_dist_ipw_rf[:, :, 0], axis=1).mean()\n", - "ground_truth_random = np.average(expected_rewards, weights=action_dist_random[:, :, 0], axis=1).mean()\n", + "policy_value_ipw_lr = np.average(expected_rewards, weights=action_dist_ipw_lr[:, :, 0], axis=1).mean()\n", + "policy_value_ipw_rf = np.average(expected_rewards, weights=action_dist_ipw_rf[:, :, 0], axis=1).mean()\n", + "policy_value_random = np.average(expected_rewards, weights=action_dist_random[:, :, 0], axis=1).mean()\n", "\n", - "print(f'policy value of IPWLearner with Logistic Regression: {ground_truth_ipw_lr}')\n", - "print(f'policy value of IPWLearner with Random Forest: {ground_truth_ipw_rf}')\n", - "print(f'policy value of Unifrom Random: {ground_truth_random}')" + "print(f'policy value of IPWLearner with Logistic Regression: {policy_value_ipw_lr}')\n", + "print(f'policy value of IPWLearner with Random Forest: {policy_value_ipw_rf}')\n", + "print(f'policy value of Unifrom Random: {policy_value_random}')" ] }, { @@ -496,7 +496,7 @@ "source": [ "In fact, IPWLearner with Random Forest reveals the best performance among the three evaluation policies.\n", "\n", - "Using the above ground-truths, we evaluate the estimation accuracy of the estimators." + "Using the above policy values, we evaluate the estimation accuracy of the OPE estimators." ] }, { @@ -522,12 +522,9 @@ "source": [ "# evaluate the estimation performances of OPE estimators \n", "# by comparing the estimated policy values of IPWLearner with Logistic Regression and its ground-truth.\n", - "# `evaluate_performance_of_estimators` returns a dictionary containing estimation performances of given estimators \n", + "# `summarize_estimators_comparison` returns a pandas dataframe containing estimation performances of given estimators \n", "relative_ee_a = ope.summarize_estimators_comparison(\n", - " ground_truth_policy_value=dataset.calc_ground_truth_policy_value(\n", - " expected_reward=bandit_feedback_test[\"expected_reward\"],\n", - " action_dist=action_dist_ipw_lr,\n", - " ),\n", + " ground_truth_policy_value=policy_value_ipw_lr,\n", " action_dist=action_dist_ipw_lr,\n", " estimated_rewards_by_reg_model=estimated_rewards_by_reg_model,\n", " metric=\"relative-ee\", # \"relative-ee\" (relative estimation error) or \"se\" (squared error)\n", @@ -560,12 +557,9 @@ "source": [ "# evaluate the estimation performance of OPE estimators \n", "# by comparing the estimated policy values of IPWLearner with Random Forest and its ground-truth.\n", - "# `evaluate_performance_of_estimators` returns a dictionary containing estimation performances of given estimators \n", + "# `summarize_estimators_comparison` returns a pandas dataframe containing estimation performances of given estimators \n", "relative_ee_b = ope.summarize_estimators_comparison(\n", - " ground_truth_policy_value=dataset.calc_ground_truth_policy_value(\n", - " expected_reward=bandit_feedback_test[\"expected_reward\"],\n", - " action_dist=action_dist_ipw_rf,\n", - " ),\n", + " ground_truth_policy_value=policy_value_ipw_rf,\n", " action_dist=action_dist_ipw_rf,\n", " estimated_rewards_by_reg_model=estimated_rewards_by_reg_model,\n", " metric=\"relative-ee\", # \"relative-ee\" (relative estimation error) or \"se\" (squared error)\n", @@ -598,12 +592,9 @@ "source": [ "# evaluate the estimation performance of OPE estimators \n", "# by comparing the estimated policy values of Uniform Random and its ground-truth.\n", - "# `evaluate_performance_of_estimators` returns a dictionary containing estimation performances of given estimators \n", + "# `summarize_estimators_comparison` returns a pandas dataframe containing estimation performances of given estimators \n", "relative_ee_c = ope.summarize_estimators_comparison(\n", - " ground_truth_policy_value=dataset.calc_ground_truth_policy_value(\n", - " expected_reward=bandit_feedback_test[\"expected_reward\"],\n", - " action_dist=action_dist_random,\n", - " ),\n", + " ground_truth_policy_value=policy_value_random,\n", " action_dist=action_dist_random,\n", " estimated_rewards_by_reg_model=estimated_rewards_by_reg_model,\n", " metric=\"relative-ee\", # \"relative-ee\" (relative estimation error) or \"se\" (squared error)\n",