Constraint learning based gradient boosting trees

作者：

Highlights：

• Two new algorithms developed, focusing on solving instance level constraint problems.

• Two new evaluation measures were developed, both instance level constraint related.

• On average, constraint measures were improved by at least 27% compared to baselines.

• On average, a minimal effect (14.7%) on the general prediction error was observed.

摘要

•Two new algorithms developed, focusing on solving instance level constraint problems.•Two new evaluation measures were developed, both instance level constraint related.•On average, constraint measures were improved by at least 27% compared to baselines.•On average, a minimal effect (14.7%) on the general prediction error was observed.

论文关键词：Machine learning,Regression,Gradient boosting,Gradient boosting trees,Constraint learning

论文评审过程：Received 12 August 2018, Revised 5 March 2019, Accepted 5 March 2019, Available online 19 March 2019, Version of Record 1 April 2019.

论文官网地址：https://doi.org/10.1016/j.eswa.2019.03.011