The Feature Importance Ranking Measure

被引:0
|
作者
Zien, Alexander [1 ,2 ]
Kraemer, Nicole [3 ]
Sonnenburg, Soeren [2 ]
Raetsch, Gunnar [2 ]
机构
[1] Fraunhofer FIRST IDA, Kekulestr 7, D-12489 Berlin, Germany
[2] Max Planck Gesell, Friedrich Miescher Lab, D-72076 Tubingen, Germany
[3] Berlin Inst Technol, Machine Learn Grp, D-10587 Berlin, Germany
关键词
VARIABLE IMPORTANCE; CLASSIFICATION; SHRINKAGE; MACHINE; MATRIX;
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Most accurate predictions are typically obtained by learning machines with complex feature spaces (as e.g. induced by kernels). Unfortunately, such decision rules are hardly accessible to humans and cannot easily be used to gain insights about the application domain. Therefore, one often resorts to linear models in combination with variable selection, thereby sacrificing some predictive power for presumptive interpretability. Here, we introduce the Feature Importance Ranking Measure (FIRM), which by retrospective analysis of arbitrary learning machines allows to achieve both excellent predictive performance and superior interpretation. In contrast to standard raw feature weighting, FIRM takes the underlying correlation structure of the features into account. Thereby, it is able to discover the most relevant features, even if their appearance in the training data is entirely prevented by noise. The desirable properties of FIRM are investigated analytically and illustrated in simulations.
引用
收藏
页码:694 / +
页数:3
相关论文
共 50 条
  • [31] A NOVEL TELECOM CUSTOMER CHURN ANALYSIS SYSTEM BASED ON RFM MODEL AND FEATURE IMPORTANCE RANKING
    Xu T.
    Ma Y.
    Ao C.
    Qu M.
    Meng X.
    Interdisciplinary Journal of Information, Knowledge, and Management, 2023, 18 : 719 - 737
  • [32] Feature importance measure of a multilayer perceptron based on the presingle-connection layer
    Wenyi Zhang
    Xiaohua Shen
    Haoran Zhang
    Zhaohui Yin
    Jiayu Sun
    Xisheng Zhang
    Lejun Zou
    Knowledge and Information Systems, 2024, 66 : 511 - 533
  • [33] Diagnosis of design-silicon timing mismatch with feature encoding and importance ranking - the methodology explained
    Bastani, Pouria
    Callegari, Nick
    Wang, Li-C.
    Abadir, Magdy S.
    2008 IEEE INTERNATIONAL TEST CONFERENCE, VOLS 1 AND 2, PROCEEDINGS, 2008, : 374 - +
  • [34] Feature importance measure of a multilayer perceptron based on the presingle-connection layer
    Zhang, Wenyi
    Shen, Xiaohua
    Zhang, Haoran
    Yin, Zhaohui
    Sun, Jiayu
    Zhang, Xisheng
    Zou, Lejun
    KNOWLEDGE AND INFORMATION SYSTEMS, 2024, 66 (01) : 511 - 533
  • [35] Ranking scholars: A measure representation
    Lasso de la Vega, Casilda
    Volij, Oscar
    JOURNAL OF INFORMETRICS, 2018, 12 (02) : 510 - 517
  • [36] Feature ranking for protein classification
    Mhamdi, F
    Rakotomalala, R
    Elloumi, M
    COMPUTER RECOGNITION SYSTEMS, PROCEEDINGS, 2005, : 611 - 617
  • [37] Wrapper for ranking feature selection
    Ruiz, R
    Aguilar-Ruiz, JS
    Riquelme, JC
    INTELLIGENT DAA ENGINEERING AND AUTOMATED LEARNING IDEAL 2004, PROCEEDINGS, 2004, 3177 : 384 - 389
  • [38] Linear feature extraction for ranking
    Pandey, Gaurav
    Ren, Zhaochun
    Wang, Shuaiqiang
    Veijalainen, Jari
    de Rijke, Maarten
    INFORMATION RETRIEVAL JOURNAL, 2018, 21 (06): : 481 - 506
  • [39] Insights into distributed feature ranking
    Bolon-Canedo, Veronica
    Sechidis, Konstantinos
    Sanchez-Marono, Noelia
    Alonso-Betanzos, Amparo
    Brown, Gavin
    INFORMATION SCIENCES, 2019, 496 : 378 - 398
  • [40] Fast feature ranking algorithm
    Ruiz, R
    Riquelme, JC
    Aguilar-Ruiz, JS
    KNOWLEDGE-BASED INTELLIGENT INFORMATION AND ENGINEERING SYSTEMS, PT 1, PROCEEDINGS, 2003, 2773 : 325 - 331