Artificial intelligence enhances maize production through smart populations within planetary boundaries
收藏资源简介:
# Dataset Description ## Raw Data Files ### 1. AIsp_origindata.xlsx- **Purpose**: Raw dataset for building the AIsp_TabPFN model- **Content**: Contains all original features and target variables used for AIsp model development---"""Soil: soil organic matter (SOM).Climate: average daily maximum temperature (Tmax), average daily minimum temperatures (Tmin), cumulative precipitation (Prep), cumulative solar radiation (Radn), and standardized precipitation–evapotranspiration index (SPEI).Management: Tillage and Irrigation.Thermal indices: growing degree days (GDD) and extreme degree days (EDD).Genotype: canopy architecture (Canopy). Target variable: optimal planting density (Dopt).Note: Optimal yield (Yopt): This variable was not utilized in the modeling process, but subsequent steps require mapping from Dopt to Yopt by canopy (smart vs. non-smart).---""" ### 2. AInm_origindata .xlsx- **Purpose**: Raw dataset for building the AInm_CatBoost model- **Content**: Contains all original features and target variables used for AInm model development---"""Soil: soil organic matter (SOM), soil total nitrogen (TotalN) and pH.Climate: average daily maximum temperature (Tmax), average daily minimum temperatures (Tmin), cumulative precipitation (Prep), cumulative solar radiation (Radn), and standardized precipitation–evapotranspiration index (SPEI).Management: Tillage and Irrigation.Thermal indices: growing degree days (GDD) and extreme degree days (EDD).Productivity: optimal yield (Yopt).Target variable: optimal nitrogen application (Nopt).---""" # Python Code Files ## 1. AIsp_compare.py- **Function**: Compares performance of different algorithms for AIsp model construction- **Process**: - Evaluates multiple machine learning algorithms - Identifies the optimal algorithm based on performance metrics - Outputs comparative analysis results ## 2. AIsp_TabPFN.py- **Function**: Implements the final AIsp model using the selected optimal algorithm- **Process**: - Performs 10-fold cross-validation with SHAP interpretation - Retrains the model using the entire dataset for final deployment - Generates model explanations and feature importance analysis ## 3. AInm_compare.py- **Function**: Compares performance of different algorithms for AInm model construction - **Process**: - Evaluates multiple machine learning algorithms - Identifies the optimal algorithm based on performance metrics - Outputs comparative analysis results ## 4. AInm_TabPFN.py- **Function**: Implements the final AInm model using the selected optimal algorithm- **Process**: - Performs 10-fold cross-validation with SHAP interpretation - Retrains the model using the entire dataset for final deployment - Generates model explanations and feature imp
# 数据集说明 ## 原始数据文件 ### 1. AIsp_origindata.xlsx - **用途**:用于构建AIsp_TabPFN(TabPFN)模型的原始数据集 - **内容**:包含用于AIsp模型开发的全部原始特征与目标变量 --- """ 土壤类:土壤有机质(soil organic matter, SOM) 气候类:日最高气温均值(average daily maximum temperature, Tmax)、日最低气温均值(average daily minimum temperature, Tmin)、累积降水量(cumulative precipitation, Prep)、累积太阳辐射量(cumulative solar radiation, Radn)以及标准化降水蒸散指数(standardized precipitation–evapotranspiration index, SPEI) 管理措施类:耕作与灌溉 热积温指标:生长度日(growing degree days, GDD)与极端度日(extreme degree days, EDD) 基因型类:冠层结构(Canopy) 目标变量:最优种植密度(optimal planting density, Dopt) 注:最优产量(optimal yield, Yopt):该变量未在建模流程中使用,但后续步骤需基于冠层类型(智能型与非智能型)建立Dopt到Yopt的映射关系。 ---""" ### 2. AInm_origindata.xlsx - **用途**:用于构建AInm_CatBoost(CatBoost)模型的原始数据集 - **内容**:包含用于AInm模型开发的全部原始特征与目标变量 --- """ 土壤类:土壤有机质(SOM)、土壤全氮(soil total nitrogen, TotalN)与pH值 气候类:日最高气温均值(Tmax)、日最低气温均值(Tmin)、累积降水量(Prep)、累积太阳辐射量(Radn)以及标准化降水蒸散指数(SPEI) 管理措施类:耕作与灌溉 热积温指标:生长度日(GDD)与极端度日(EDD) 生产力类:最优产量(Yopt) 目标变量:最优氮肥施用量(optimal nitrogen application, Nopt) ---""" ## Python代码文件 ### 1. AIsp_compare.py - **功能**:对比不同算法在构建AIsp模型时的性能表现 - **流程**: - 评估多种机器学习算法 - 基于性能指标筛选最优算法 - 输出对比分析结果 ### 2. AIsp_TabPFN.py - **功能**:使用筛选出的最优算法实现最终的AIsp_TabPFN模型 - **流程**: - 结合SHAP(SHapley Additive exPlanations)解释进行10折交叉验证 - 使用全量数据集重新训练模型以用于最终部署 - 生成模型解释与特征重要性分析结果 ### 3. AInm_compare.py - **功能**:对比不同算法在构建AInm模型时的性能表现 - **流程**: - 评估多种机器学习算法 - 基于性能指标筛选最优算法 - 输出对比分析结果 ### 4. AInm_TabPFN.py - **功能**:使用筛选出的最优算法实现最终的AInm模型 - **流程**: - 结合SHAP(SHapley Additive exPlanations)解释进行10折交叉验证 - 使用全量数据集重新训练模型以用于最终部署 - 生成模型解释与特征重要性分析结果



