In Machine learning and Statistics, feature selection (특징 선택), also known as variable selection, attribute selection or variable subset selection, is the process of selecting a subset of relevant features (variables, predictors) for use in model construction. (https://en.wikipedia.org/wiki/Feature_selection)

특징 선택은 모델에서 중요한 특징을 선택하는 것이고, 특징 추출(Feature extraction)은 임의의 특징을 만들어 내는 것.

모델을 만들기전에 Feature selection을 수행하면 좋은 점

1. Reduce overfitting: 중복 데이터를 줄인다.
2. Improves accuracy: 정확도 향상
3. Reduces training time: 학습시간을 줄인다.

방법들

1. Univariate selection: 각 특징별 빈도로 Chi-squared test 하여, 유의한 특징 선별
2. Recursive feature elimination: 재귀적으로 특징 제거
3. PCA weights
4. 일부 기계학습 알고리즘에서 important features 정보 제공 (Random forest, Extra trees classifiers,...)

관련정보

• Beginner's Guide to Feature Selection in Python 2018-09-26
• Feature Selection For Machine Learning in Python 2016-05-20: Using Python
• Principal Component Analysis (PCA) for Feature Selection and some of its Pitfalls 2016-03-24: PCA로 특징 선택
• 차원축소와 특징선택이란 무엇인가 2016-05-05

관련논문

• Feature selecction methods for big data bioinformatics: A survey from the search perspective 2016-12-??
• Feature Selection: A Data Perspective 2016-01-29
• 의미 기반 유전 알고리즘을 사용한 특징 선택 Journal of Korean Socieity for Internet Information 2012-08-??
• An Introduction to Variable and Feature Selection Journal of Machine learning Research 2013-03-03

Suggested Pages #

• 0.025 July 19
• 0.025 Programming language
• 0.025 NumPy
• 0.025 Matplotlib
• 0.025 January 17
• 0.025 December 1
• 0.025 November 11
• 0.025 Pattern recognition
• 0.025 207
• 0.025 TinyDB
• More suggestions...