Off-Policy Evaluation

Open Bandit Dataset and Pipeline: Towards Realistic and Reproducible Off-Policy Evaluation
Off-policy evaluation (OPE) aims to estimate the performance of hypothetical policies using data generated by a different policy. …
Yuta Saito, Shunsuke Aihara, Megumi Matsutani, Yusuke Narita
Cite Code Dataset Proceedings arXiv
Counterfactual Learning and Evaluation for Recommender Systems
Counterfactual estimators enable the use of existing log data to estimate how some new target recommendation policy would have …
Yuta Saito, Thorsten Joachims
Cite Code Video Proceedings Website
Evaluating the Robustness of Off-Policy Evaluation
Off-policy Evaluation (OPE), or offline evaluation in general, evaluates the performance of hypothetical policies leveraging only …
Yuta Saito, Takuma Udagawa, Haruka Kiyohara, Kazuki Mogi, Yusuke Narita, Kei Tateno
Cite Code Slides Proceedings arXiv
Optimal Off-Policy Evaluation from Multiple Logging Policies
We study off-policy evaluation (OPE) from multiple logging policies, each generating a dataset of fixed size, i.e., stratified …
Nathan Kallus, Yuta Saito, Masatoshi Uehara
Cite Code Proceedings arXiv
Doubly Robust Estimator for Ranking Metrics with Post-Click Conversions
Post-click conversion, a pre-defined action on a web service after a click, is an essential form of feedback, as it directly …
Yuta Saito
Cite Code Slides Proceedings
Data-Driven Off-Policy Estimator Selection: An Application in User Marketing on An Online Content Delivery Service
Off-policy evaluation (OPE) is the method that attempts to estimate the performance of decision making policies using historical data …
Yuta Saito, Takuma Udagawa, Kei Tateno
Cite arXiv
Open Bandit Pipeline
A research framework for off-policy evaluation and learning
Last updated on Feb 2, 2022
PDF