基本信息:
- 专利标题： 선형 및 비선형 학습기법을 이용한 정량적 구조-활성 상관관계 기반 녹는점 예측을 위한 물성예측모델
- 专利标题（英）：Melting-Point Prediction Model based on Quantitive Structure-Activity Relationships with Linear and Non linear Machine Learning Methods
- 专利标题（中）：- 基于线性和非线性机器学习方法的量化结构 - 活动关系的熔点预测模型
- 申请号：KR1020150058957 申请日：2015-04-27
- 公开(公告)号：KR1020160127486A 公开(公告)日：2016-11-04
- 发明人： 윤정혁 , 장병하 , 김한조 , 서영주 , 정우성 , 강경태 , 노경태 , 김광연 , 신성은
- 申请人： 주식회사 크레아플래닛
- 申请人地址： *-***, **, Beonnyeong-ro, Ansan-si danwon-gu, Gyeonggi-do, (***-***), Republic of Korea
- 专利权人： 주식회사 크레아플래닛
- 当前专利权人： 주식회사 크레아플래닛
- 当前专利权人地址： *-***, **, Beonnyeong-ro, Ansan-si danwon-gu, Gyeonggi-do, (***-***), Republic of Korea
- 代理人： 특허법인세원
- 主分类号： G06F19/00
- IPC分类号： G06F19/00 ; G01N33/00
The present invention is quantitative structure using the linear and non-linear learning techniques - as based on the activity correlation on the Properties prediction model that can be simply and accurately predict the melting point of the chemical, physical properties predicted for the melting point prediction according to the invention method, 1) collecting a purified compound having a melting point data value; 2) collecting, calculates the expression to calculate the molecular structure of the compound data; 3) dividing the collected to give the above compound as a training data set and the external validation set; 4) step forward to the presenter's first filtration operation; 5) selecting the appropriate characters represented in boiling gender prediction; 6) Each machine learning model optimized increase characters expressed and accumulate the result and the final model through each model two or more of a combination of models of the calculated generate a combined model to reduce the error occurring in each model by intercomparison determining a; 7) setting the applicable range for the reliability evaluation in the final model and apply; 8) determining the authenticity and suitability of the final model; includes.