Monthly Dacon Bio-optical Data Analysis Contest

Algorithm | Structured | Regression | Bio | MAE

Prize : Total 360만원
978 Users Completed

Practice

Overview Data Code (share) Talk Leaderboard

Submission

Private LB : 0.92553 (Rank 17)

Jamm

2020.06.27 00:37 9,391 Views language

제가 올렸던 LB 1.09 글에서 조금씩 변수선택, 모델링, 스태킹앙상블 한 코드입니다. 더이상 뭘 해볼까 생각이 나지 않아서 쓸데없이 스태킹 앙상블 하려고 시간 쓰고있었는데, 다른 분들 솔루션 보니까, 제가 전혀 눈치채지 못한 데이터의 특성, 저는 생각도 해보지 않은 부분의 변수 만들기... 아직 배울것이 참 많네요.
역시 적절한 피쳐엔지니어링은 최고의 결과로 항상 보답하는군요. Good Models, Good Features! 피쳐엔지니어링의 중요성에 대해서 새삼 느끼고 갑니다. 모두들 고생 많으셨습니다!

Code

4 Comments

comment

0 / 1000

2020.06.27 00:45

Wow! It's pretty complex! Good job! It's very interesting to see the DFT here. Is it really useful?

Jamm

2020.06.27 00:51

It was, but not very much. My 10 Fold CV Score just got better from 1.12 to 1.09.
Building Genetic Algorithm for Feature selection, and tuning and managing 12 (4 targets, 3 algs) model was tough job for me. Congrats on winning! Your solution was very impressive... :)

2020.06.27 00:56

Thanks! Also want to thank you one more time for your very insightful post about spectrometry. It was a very good start point to work without any knowledge about the subject. Absorption... yeah.

23fdf51d

2020.06.27 01:22

공유 감사합니다. 도움이 많이 될 것 같네요!!

📣 The comment input field has been moved to the top of the comment list!

List

코드 공유 (LB: 0.85873)

Competition - 월간 데이콘 생체 광학 데이터 분석 AI 경진대회

Current