The parameter sensitivity of random forests.

The parameter sensitivity of random forests.
复制标题

DOI:
10.1186/s12859-016-1228-x
复制
发表时间:
2016-09-01
期刊:
影响因子:
3
通讯作者:
Boutros PC
Boutros PC
中科院分区:
生物学4区
文献类型:
--
作者:
Huang BF;Boutros PC

文献摘要

参考文献

被引文献

相似文献

用于监督机器学习的随机森林(RF)算法是一种广泛应用于科学和许多其他领域的集成学习方法。它的受欢迎程度一直在增加,但相对较少的研究涉及参数选择过程:模型拟合的关键步骤。由于有关默认参数性能可靠性的大量断言,许多 RF 模型都适合使用这些值。然而,在计算基因组研究中尚未对 RF 的参数敏感性进行彻底的检查。我们在这里解决这个差距。我们使用 RF 机器学习算法在两个具有不同 p/n 比的生物数据集上检查了参数选择对分类性能的影响:测序摘要统计数据(低 p/n)和微阵列衍生数据(高 p/n)。这里,p 是指变量的数量,n 是样本的数量。我们的研究结果表明,参数化与预测准确性和变量重要性度量(VIM)高度相关。此外,我们证明了不同的参数对于调整不同的数据集至关重要,并且参数优化显着增强了默认参数。参数性能在低和高 p/n 数据上表现出很大的可变性。因此,通过模型调整 RF 使其远离其默认参数设置可以获得显着的好处。本文的在线版本 (doi:10.1186/s12859-016-1228-x) 包含补充材料,可供授权用户使用。
The Random Forest (RF) algorithm for supervised machine learning is an ensemble learning method widely used in science and many other fields. Its popularity has been increasing, but relatively few studies address the parameter selection process: a critical step in model fitting. Due to numerous assertions regarding the performance reliability of the default parameters, many RF models are fit using these values. However there has not yet been a thorough examination of the parameter-sensitivity of RFs in computational genomic studies. We address this gap here. We examined the effects of parameter selection on classification performance using the RF machine learning algorithm on two biological datasets with distinct p/n ratios: sequencing summary statistics (low p/n) and microarray-derived data (high p/n). Here, p, refers to the number of variables and, n, the number of samples. Our findings demonstrate that parameterization is highly correlated with prediction accuracy and variable importance measures (VIMs). Further, we demonstrate that different parameters are critical in tuning different datasets, and that parameter-optimization significantly enhances upon the default parameters. Parameter performance demonstrated wide variability on both low and high p/n data. Therefore, there is significant benefit to be gained by model tuning RFs away from their default parameter settings. The online version of this article (doi:10.1186/s12859-016-1228-x) contains supplementary material, which is available to authorized users.
DOI: 10.1186/s12859-015-0787-6
发表时间: 2015-10-28
期刊: BMC bioinformatics
影响因子: 3
作者:
Cremona MA;Sangalli LM;Vantini S;Dellino GI;Pelicci PG;Secchi P;Riva L
通讯作者: Riva L
通过基于局部超平面的学习策略来优化特征选择和分类的组合
DOI: 10.1186/s12859-015-0629-6
发表时间: 2015-07-10
期刊: BMC bioinformatics
影响因子: 3
作者:
Cheng X;Cai H;Zhang Y;Xu B;Su W
通讯作者: Su W
DOI: 10.1016/j.csda.2007.08.015
发表时间: 2008-01-10
影响因子: 1.8
作者:
Archer, Kelfie J.;Kirnes, Ryan V.
通讯作者: Kirnes, Ryan V.
DOI: 10.1023/a:1007607513941
发表时间: 2000-08-01
期刊: MACHINE LEARNING
影响因子: 7.5
作者:
Dietterich, TG
通讯作者: Dietterich, TG
DOI: 10.1186/s12859-015-0723-9
发表时间: 2015-09-21
期刊: BMC bioinformatics
影响因子: 3
作者:
Blagus R;Lusa L
通讯作者: Lusa L