hrcak mascot   Srce   HID

Izvorni znanstveni članak
https://doi.org/10.2478/bsrj-2018-0017

Number of Instances for Reliable Feature Ranking in a Given Problem

Marko Bohanec   ORCID icon orcid.org/0000-0002-5295-5111 ; Salvirt Ltd., Ljubljana, Slovenia
Mirjana Kljajić Borštnar   ORCID icon orcid.org/0000-0003-4608-9090 ; Faculty of Organizational Sciences, University of Maribor, Kranj, Slovenia
Marko Robnik-Šikonja   ORCID icon orcid.org/0000-0002-1232-3320 ; Faculty of Computer and Information Science, University of Ljubljana, Ljubljana, Slovenia

Puni tekst: engleski, pdf (528 KB) str. 35-44 preuzimanja: 124* citiraj
APA 6th Edition
Bohanec, M., Kljajić Borštnar, M. i Robnik-Šikonja, M. (2018). Number of Instances for Reliable Feature Ranking in a Given Problem. Business Systems Research, 9 (2), 35-44. https://doi.org/10.2478/bsrj-2018-0017
MLA 8th Edition
Bohanec, Marko, et al. "Number of Instances for Reliable Feature Ranking in a Given Problem." Business Systems Research, vol. 9, br. 2, 2018, str. 35-44. https://doi.org/10.2478/bsrj-2018-0017. Citirano 16.07.2019.
Chicago 17th Edition
Bohanec, Marko, Mirjana Kljajić Borštnar i Marko Robnik-Šikonja. "Number of Instances for Reliable Feature Ranking in a Given Problem." Business Systems Research 9, br. 2 (2018): 35-44. https://doi.org/10.2478/bsrj-2018-0017
Harvard
Bohanec, M., Kljajić Borštnar, M., i Robnik-Šikonja, M. (2018). 'Number of Instances for Reliable Feature Ranking in a Given Problem', Business Systems Research, 9(2), str. 35-44. doi: https://doi.org/10.2478/bsrj-2018-0017
Vancouver
Bohanec M, Kljajić Borštnar M, Robnik-Šikonja M. Number of Instances for Reliable Feature Ranking in a Given Problem. Business Systems Research [Internet]. 2018 [pristupljeno 16.07.2019.];9(2):35-44. doi: https://doi.org/10.2478/bsrj-2018-0017
IEEE
M. Bohanec, M. Kljajić Borštnar i M. Robnik-Šikonja, "Number of Instances for Reliable Feature Ranking in a Given Problem", Business Systems Research, vol.9, br. 2, str. 35-44, 2018. [Online]. doi: https://doi.org/10.2478/bsrj-2018-0017

Sažetak
Background: In practical use of machine learning models, users may add new features to an existing classification model, reflecting their (changed) empirical understanding of a field. New features potentially increase classification accuracy of the model or improve its interpretability. Objectives: We have introduced a guideline for determination of the sample size needed to reliably estimate the impact of a new feature. Methods/Approach: Our approach is based on the feature evaluation measure ReliefF and the bootstrap-based estimation of confidence intervals for feature ranks. Results: We test our approach using real world qualitative business-to-business sales forecasting data and two UCI data sets, one with missing values. The results show that new features with a high or a low rank can be detected using a relatively small number of instances, but features ranked near the border of useful features need larger samples to determine their impact. Conclusions: A combination of the feature evaluation measure ReliefF and the bootstrap-based estimation of confidence intervals can be used to reliably estimate the impact of a new feature in a given problem.

Ključne riječi
machine learning; feature ranking; feature evaluation

Hrčak ID: 203480

URI
https://hrcak.srce.hr/203480

Posjeta: 190 *