본 연구는 환경 요인을 바탕으로 절화용 국화 생장 예측을 위한 최적의 모델을 개발하는 것을 목표로 하였다. 이를 위해 13개의 모델(Linear Regression, Lasso Regression, Ridge Regression, ElasticNet Regression, K-Nearest Neighbors (KNN), Support Vector Regression (SVR), Neural Network, Decision Tree, Random Forest, XGBoost, AdaBoost, CatBoost, Stacking)의 성능을 R2, MAE, RMSE를 평가 지표 로 비교하였다. 단일 모델 중에서는 Decision Tree가 가장 우수한 성능을 보였으며, R2값은 0.90에서 0.91 사이였다. 앙 상블 모델 중에서는 CatBoost가 가장 높은 성능을 보였으며 (R2=0.90~0.92) Random Forest와 XGBoost 또한 유사한 성 능을 보였다. 전체적으로 트리 기반 앙상블 모델이 국화 생장 예측에 적합한 모델로 나타났다.

To find an alternative for synthetic pesticides, methanol extract from plant samples were tested for their insecticidal activity against insect. The extract of Asiasarum sieboldii had strongly insecticidal activity against Plutella xylostella. Roots of A. sieboldii were extracted with methanol, and the concentrated extract was partitioned with n-hexane, ethylacetate, n-buthanol and H2O. The highest activity was shown in the hexane fraction. Activity-guided fractionation led to the isolation of two amides from hexane fration through the repeated silica gel column chromatographic separations. From the interpretation of spectropic data including NMR, MS, IR, the chemical structures of compounds were determined as dodeca-2E,4E,8Z,10Z-tetraenoic acid isobutylamide and dodeca-2E,4E,8Z, 10E-tetraenoic acid isobutylamide. These compounds showed insecticidal activity on P. xylostella by 96.7% at 100ppm. The liquid formulation controlled on cabbage effectively. The extract and compounds from A. sieboldii showed insecticidal activity against Nilaparvata lugens. As a naturally occurring pesticide, A. sieboldii could be useful as a new botanic insecticide.