tesseract-research/atlas-global

Atlas v1.15.0是一个统一、标准化且生产就绪的数据集，旨在整合全球多个国家的经济和生产数据。该数据集包含81,229,968行数据，总大小为414.2 MB，由88个文件组成。核心数据文件包括：cells_master_global.parquet（419,895行，统一的生产单元数据，已进行NAICS/ISIC/NACE行业分类交叉映射、美元标准化和质量评分）、eu_cells_percentiles.parquet（18,132行，提供欧盟国家按国家、NACE行业和年份划分的收入、就业和工资的百分位分布p10/p25/p50/p75/p90）、industry_crosswalk.parquet（26行，提供NACE章节与ISIC和NAICS-2位码的映射关系）。数据覆盖40多个国家，包括美国、加拿大、澳大利亚、欧盟成员国（如德国、法国）、日本、巴西、新加坡等，数据来源包括Eurostat、国家统计局（如美国人口普查局、日本e-Stat）和公开API（如阿根廷datos.gob.ar）。数据集经过处理，包括行业交叉映射、外汇汇率转换、质量重新校准和美元标准化，以支持生产环境查询。此外，数据集还包含原始国家级数据文件（如7个per-country cells_master.parquet文件）和欧盟百分位插值数据，用于分析企业级分布。该版本重点关注“统一程序方面”，确保数据模式一致，并优化了存储架构（如Supabase Pro加载）。数据集适用于经济分析、市场研究和SaaS应用查询。Atlas v1.15.0 is a unified, standardized, and production-ready dataset designed to integrate economic and production data from multiple countries worldwide. The dataset contains 81,229,968 rows with a total size of 414.2 MB, comprising 88 files. Key data files include: cells_master_global.parquet (419,895 rows, unified production cells with NAICS/ISIC/NACE industry crosswalking, USD normalization, and quality scoring), eu_cells_percentiles.parquet (18,132 rows, providing percentile distributions p10/p25/p50/p75/p90 of revenue, employment, and payroll per country, NACE industry, and year for the EU), and industry_crosswalk.parquet (26 rows, mapping NACE sections to ISIC and NAICS-2 digits). Data coverage spans over 40 countries, including the United States, Canada, Australia, EU member states (e.g., Germany, France), Japan, Brazil, Singapore, and others, sourced from Eurostat, national statistical offices (e.g., U.S. Census Bureau, Japans e-Stat), and public APIs (e.g., Argentinas datos.gob.ar). The dataset has been processed with industry crosswalking, foreign exchange rate conversion, quality recalibration, and USD normalization to support production environment queries. Additionally, it includes raw country-level data files (e.g., 7 per-country cells_master.parquet files) and EU bucket interpolation data for firm-level distribution analysis. This version focuses on the unified procedural aspect, ensuring consistent schema and optimized storage architecture (e.g., Supabase Pro loading). The dataset is suitable for economic analysis, market research, and SaaS application queries.