3.Summary - 3. 빅데이터 모델 평가 검증하기

 

 

 

과적합(overfitting)

·         과적합은 제한된 샘플(혹은 훈련에 사용한 한정된 데이터)에 너무 특화가 되어, 새로운 샘플에 대한 예측의 결과가 오히려 나빠지거나 오차가 매우 커지거나 학습의 효과가 나타나지 않는 경우를 말합니다.

·         비학습 데이터 혹은 향후에 만들어질 모델에 대해 예측력이 떨어지거나 성능이 좋지 않은 상태를 의미합니다.

 

https://codedragon.tistory.com/8362

https://codedragon.tistory.com/8416

https://codedragon.tistory.com/10000

https://codedragon.tistory.com/9763

https://codedragon.tistory.com/9456