Performance evaluation of outlier detection techniques in production timeseries: A systematic review and meta-analysis
作者：
Highlights：
• Evaluated 17 algorithms for outlier detection in oil/gas production timeseries.
• Compared the performance of the 17 algorithms based on 6 different metrics.
• Selected the top 8 algorithms based on the examination outcome of synthetic data.
• Compared the performance of the top 8 techniques on two real datasets.
• KNN is a simple yet the top technique that can handle different production trends.
摘要
•Evaluated 17 algorithms for outlier detection in oil/gas production timeseries.•Compared the performance of the 17 algorithms based on 6 different metrics.•Selected the top 8 algorithms based on the examination outcome of synthetic data.•Compared the performance of the top 8 techniques on two real datasets.•KNN is a simple yet the top technique that can handle different production trends.
论文关键词：Outlier detection,Production data analysis,Decline curve analysis,Performance evaluation,Binary classification
论文评审过程：Received 22 April 2021, Revised 29 November 2021, Accepted 30 November 2021, Available online 4 December 2021, Version of Record 8 December 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.116371
