Optimized Risk Scores

Optimized Risk Scores
复制标题

优化的风险评分

DOI:
10.1145/3097983.3098161
复制
发表时间:
2017
期刊:
Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining
影响因子:
--
通讯作者:
C. Rudin
C. Rudin
中科院分区:
--
文献类型:
--
作者:
Berk Ustun;C. Rudin

文献摘要

参考文献

被引文献

相似文献

风险分数是简单的分类模型,让用户通过添加,减去和乘以几个小数字来快速评估风险。这种模型广泛用于医疗保健和刑事司法,但通常是临时建立的。在本文中,我们提出了一种原则性的方法来学习风险分数,这些分数针对特征选择、整数系数和操作约束进行了充分优化。我们将风险评分问题转化为一个混合整数非线性规划,并提出了一种新的割平面算法来有效地恢复其最优解。我们的方法可以以一种与数据集的样本大小线性缩放的方式拟合优化的风险分数,提供最优性证明,并且在没有参数调整的情况下遵守复杂的约束。我们通过一组广泛的数值实验和一个应用程序来说明这些好处,在这个应用程序中,我们为ICU癫痫发作预测建立了一个定制的风险评分。
Risk scores are simple classification models that let users quickly assess risk by adding, subtracting, and multiplying a few small numbers. Such models are widely used in healthcare and criminal justice, but are often built ad hoc. In this paper, we present a principled approach to learn risk scores that are fully optimized for feature selection, integer coefficients, and operational constraints. We formulate the risk score problem as a mixed integer nonlinear program, and present a new cutting plane algorithm to efficiently recover its optimal solution. Our approach can fit optimized risk scores in a way that scales linearly with the sample size of a dataset, provides a proof of optimality, and obeys complex constraints without parameter tuning. We illustrate these benefits through an extensive set of numerical experiments, and an application where we build a customized risk score for ICU seizure prediction.
DOI: 10.1016/s1062-1458(01)00458-5
发表时间: 2001-11
期刊: JAMA
影响因子: --
作者:
B. Gage;A. Waterman;W. Shannon;M. Boechler;M. Rich;M. Radford
通讯作者: B. Gage;A. Waterman;W. Shannon;M. Boechler;M. Rich;M. Radford
DOI: 10.18637/jss.v033.i01
发表时间: 2010-02-01
影响因子: 5.8
作者:
Friedman, Jerome;Hastie, Trevor;Tibshirani, Rob
通讯作者: Tibshirani, Rob
DOI: 10.5664/jcsm.5476
发表时间: 2016-01-01
影响因子: 4.3
作者:
Ustun, Berk;Westover, Brandon;Bianchi, Matt T.
通讯作者: Bianchi, Matt T.