Raw ESG data of 10,000+ listed companies in Europe, including scope 1, 2 & 3

# Introduction

This carbon emissions dataset provides an overview of the scope 1, 2, and 3 emissions of publicly listed companies in Europe. Oxari ensures full data coverage across all three scopes by leveraging AI to model emission profiles of companies that partly report or lack reporting at all. Our AI based methodology incorporates various factors to precisely model emission data. These factors include for example industry specifics, balance sheet data and financial statements.

# Our approach

Unlike traditional approaches that rely solely on directly reported data and pre-existing databases, we leverage the latest developments in AI and machine learning to model corporate carbon emissions when reporting is lacking. More information about our model can be found in the linked documentation. Following this approach we provide access to reported and modeled emission data for over 10,000 companies.

# Use cases

- **Historic emissions modeling:** With our in house developed AI models we calculated historic emissions of individual companies and portfolios.
- **Dataset completion:** Our AI models enable you to fill gaps in emission datasets effortlessly. For example, if scope 3 is missing, we model the expected scope 3 value and include it in the data ensuring 100% data coverage.
- **Carbon footprint assessment:** With our data you can assess the environmental impact of any investment portfolio by quantifying carbon emissions associated with each holding.
- **Identify environmental leaders and laggards:** Oxari facilitates informed decision making, since you can identify the frontrunners and laggards in emission reduction efforts.
- **Decoupling evaluation:** You can evaluate the decoupling scale of companies by analyzing their economic growth pathway and compare it to their climate impact and emissions.

# AI pipeline

Our AI pipeline consists of five steps, starting with data collection and cleaning all the to emission modeling. The process is as follows:

1. **Data collection:** Oxari gathers granular emission data from trusted sources, such as the OECD and company reports.
2. **Data cleaning:** We validate the collected datapoints both manually and automatically to ensure data integrity.
3. **Categorize scope data:** A cleaned emission dataset is categorized into variables, such as scope 1, 2, and 3. Collected financial data, such as income statements and industry specific data is categorized and added to the datasets.
4. **AI modeling and training:** Oxari developed multiple AI models that identify distinct patterns in the datasets. We use these academically validated models to compute missing carbon emissions and ensure full data coverage.
5. **Result:** Our AI models compute missing emissions associated with a specific company, based on the company's unique profile. This output is derived from a weighted average of variables in our models, tailored towards the unique characteristics of a company, such as the sector and industry.

# Other

Our dataset supports sustainability disclosures, ESG reporting, and regulatory compliance initiatives. Hedge funds, venture capital, banks, impact consultants, and insurance companies leverage our data to evaluate investment opportunities and identify potential growth prospects across various industries.

Included are direct emissions (Scope 1), indirect emissions (Scope 2), and upstream and downstream emissions (Scope 3).

# 引言
本碳排放数据集概览了欧洲公开上市企业的范围1、范围2及范围3（Scope 1/2/3）碳排放情况。Oxari通过运用人工智能对部分披露或完全未披露碳排放信息的企业构建排放画像，实现了三大核算范围的全数据覆盖。我们基于人工智能的建模方法整合多维度因素以精准构建碳排放数据模型，这些因素包括但不限于行业特性、资产负债表数据及财务报表。

# 建模方法
与仅依赖直接披露数据及既有数据库的传统方法不同，我们运用人工智能与机器学习领域的最新进展，为未披露碳排放信息的企业构建碳排放模型。有关本模型的更多细节可查阅附带的文档。通过该方法，我们可为超1万家企业提供已披露及经建模生成的碳排放数据。

# 应用场景
- **历史碳排放建模：** 依托自研人工智能模型，我们可计算单个企业及投资组合的历史碳排放数据。
- **数据集补全：** 我们的人工智能模型可轻松填补碳排放数据集的缺失项。例如，若某企业未披露范围3碳排放数据，我们将建模生成预期的范围3碳排放值并纳入数据集，实现100%数据覆盖。
- **碳足迹评估：** 借助本数据集，您可通过量化每一笔持仓对应的碳排放量，评估任意投资组合的环境影响。
- **识别环保领跑者与落后者：** Oxari可助力您精准识别在减排工作中处于领先地位的企业及滞后企业，辅助科学决策。
- **脱钩评估：** 您可通过分析企业的经济增长路径，并将其与气候影响及碳排放水平进行对比，评估企业的脱碳脱钩程度。

# 人工智能处理流程
我们的人工智能处理流程包含五个步骤，从数据采集、清洗直至碳排放建模，具体流程如下：
1. **数据采集：** Oxari从经合组织（OECD）、企业报告等可信来源获取精细化碳排放数据。
2. **数据清洗：** 我们通过人工与自动结合的方式对采集到的数据点进行校验，以保障数据完整性。
3. **碳排放范围分类：** 将清洗完成的碳排放数据集按变量分类，例如范围1、范围2及范围3；同时将采集到的财务数据（如利润表）及行业专属数据进行分类并纳入数据集。
4. **人工智能建模与训练：** Oxari开发了多款人工智能模型，可识别数据集中的差异化模式。我们采用经学术验证的此类模型计算缺失的碳排放数据，实现全数据覆盖。
5. **结果输出：** 我们的人工智能模型基于企业的独特画像，计算其缺失的碳排放数据。该输出结果由模型中各变量的加权平均生成，并针对企业的板块、行业等独特特征进行定制化调整。

# 其他说明
本数据集可支撑可持续信息披露、ESG报告及监管合规相关工作。对冲基金、风险投资机构、银行、影响力咨询机构及保险公司均可借助本数据集评估投资机会，挖掘各行业的潜在增长前景。
本数据集涵盖直接碳排放（范围1，Scope 1）、间接碳排放（范围2，Scope 2）以及上下游碳排放（范围3，Scope 3）。

搜集汇总

背景与挑战

背景概述

该数据集涵盖欧洲10,000余家上市公司的范围1/2/3碳排放原始数据，通过AI建模技术填补未报告或部分报告企业的排放数据缺口，支持碳足迹评估、ESG披露及投资组合分析等应用场景。