sweetapricity/bridgedeck-nbi

Bridgedeck — 国家桥梁清单（2000–2023）是bridgedeck数据集集合的一部分，专注于美国桥面板劣化研究。该数据集包含2000年至2023年美国所有州（包括华盛顿特区和波多黎各）清理和解析后的国家桥梁清单（NBI）记录，每条记录对应每座桥梁在检查年份的数据。数据集覆盖桥梁的结构属性、几何特征、材料信息、交通流量、状况评级，并衍生出一个5年桥面板劣化目标变量。NBI作为核心数据集，其他相关数据集（如PRISM、NOAA、NSHM、NFHL、Salt）通过bridge_id或county_fips与之连接。源数据来自FHWA国家桥梁清单，共1,248个原始.txt文件。数据模式包括标识符和位置、设计和材料、操作参数、状况评级（0-9分制，9为优秀）以及衍生特征。机器学习目标为deck_drop_5yr，是一个二分类变量，表示桥面板状况评级在未来5年内是否下降至少1点。数据经过质量处理，剔除无效坐标和重复记录。该数据集适用于土木工程、基础设施管理和预测建模研究。Bridgedeck — National Bridge Inventory (2000–2023) is part of the bridgedeck dataset collection for US bridge deck deterioration research. It contains cleaned and parsed National Bridge Inventory (NBI) records for all US states, including Washington DC and Puerto Rico, from 2000 to 2023, with one row per bridge per inspection year. The dataset covers structural properties, geometry, materials, traffic, condition ratings, and a derived 5-year deck deterioration target. The NBI serves as the backbone of the bridgedeck dataset, with other datasets (PRISM, NOAA, NSHM, NFHL, Salt) joined to it via bridge_id or county_fips. Source data is from the FHWA National Bridge Inventory, comprising 1,248 raw .txt files. The schema includes identifiers and location, design and material, operational features, condition ratings (on a 0–9 scale, where 9 is excellent), and derived features. The machine learning target is deck_drop_5yr, a binary classification variable indicating whether the bridges deck condition rating dropped by at least 1 point within the next 5 years. Data quality measures include dropping zero-encoded coordinates and deduplicating records. This dataset is suitable for civil engineering, infrastructure management, and predictive modeling research.