| Autor: |
Ghazali, Nurul Bashirah, Ramli, Khairun Nidzam, Seman, Fauziahanim Che, Isa, Khalid, Abidin, Zuhairiah Zainal, Mustam, Saizalmursidi Md, Bakar, Nurul Huda A., Haek, Mohammed Al |
| Předmět: |
|
| Zdroj: |
AIP Conference Proceedings; 2023, Vol. 2564 Issue 1, p1-11, 11p |
| Abstrakt: |
Training machine learning model may involve a huge amount of data with a variety of variables, be it numerical variable or categorical variable. Using all the variables in a dataset may cause model complexity and could potentially reduce accuracy. Concerning the possibility of accuracy reduction related to the number of variables, one of the methods introduced that aids in model's efficiency is feature selection which is one of the feature space dimensionality reduction method. Feature selection aids in selection variables that contributes to the prediction variables. Therefore, this paper compared several methods such as ANOVA, Random Forest Feature Importance, Mutual Information Feature Selection and Univariate Regression Test, in feature selection that successfully rank the important parameters that suits the numerical and categorical target of VDSL2 Technology Network dataset. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
