Raw Datasets for I-AMPPM Framework Validation: Multi-Scenario ROC Metrics and 90-Run Simulation Results

This repository contains the raw empirical data and performance validation metrics supporting the research article: "An Intelligent and Adaptive Environmental Governance Framework for Sustainable Oceans: Policy-Oriented Integration of Monitoring and Response Systems for Ports and Offshore Pollution." To ensure full scientific transparency and reproducibility, we have provided the following two primary datasets: 1. Supplementary Data Sheet 1: ROC Analysis Metrics (roc_data_combined.csv) This dataset contains the performance evaluation results for the AI-driven sensing layer across three critical maritime pollution scenarios: Port-area Oil Spill, Offshore Vessel Leakage, and Illegal Discharge. Metrics included: False Positive Rate (FPR), True Positive Rate (TPR), Youden’s J Index, and Area Under the Curve (AUC). 2. Supplementary Data Sheet 2: Simulation Raw Dataset (I-AMPPM_simulation_dataset.csv) This dataset provides the complete output records for 90 independent simulation runs (30 runs per scenario). Variables included: run_id, scenario type, area_true_m2, area_pred_m2, latency_ms, precision, recall, and RMSE. Methodological Summary: The data was generated using a high-fidelity maritime simulation environment. AI detection was validated against varied environmental noise. All statistical analyses and 95% confidence intervals reported in the study were derived from these raw datasets.本仓库包含支撑研究论文《面向海洋可持续发展的智能自适应海洋治理框架：面向港口与近海污染的监测与响应系统的政策导向型集成》的原始实验数据与性能验证指标。 为确保充分的科学透明度与研究可复现性，我们提供了以下两大核心数据集： 1. 补充数据表1：ROC分析指标集（roc_data_combined.csv）。该数据集涵盖AI感知层在三类关键海洋污染场景下的性能评估结果，分别为港口溢油、近海船舶泄漏与非法排污。包含的评估指标包括：假阳性率（False Positive Rate, FPR）、真阳性率（True Positive Rate, TPR）、约登J指数（Youden’s J Index）以及曲线下面积（Area Under the Curve, AUC）。 2. 补充数据表2：仿真原始数据集（I-AMPPM_simulation_dataset.csv）。该数据集包含90次独立仿真运行的完整输出记录（每个场景对应30次运行）。所涉变量包括：运行ID（run_id）、场景类型、真实面积（area_true_m2）、预测面积（area_pred_m2）、延迟时长（latency_ms）、精确率（precision）、召回率（recall）以及均方根误差（Root Mean Square Error, RMSE）。 方法学概述：本数据集通过高保真海洋仿真环境生成。AI检测模型已针对多样化的环境噪声条件完成验证。本研究中报告的全部统计分析结果与95%置信区间均源自上述原始数据集。