Overview
Sirma leverages advanced machine learning (ML) technologies to tackle a significant challenge in environmental sustainability: accurately estimating Scope 3 carbon emissions. They encompass indirect emissions generated throughout the value chain, such as those from logistics and shipping, and are notoriously difficult to measure due to the complexities and variety of data sources involved. Sirma’s ML system effectively classifies shipment data, delivering detailed, reliable, and actionable estimates of carbon emissions. This valuable information assists organizations in achieving their sustainability goals and fulfilling regulatory reporting requirements.
The Challenge
Accounting for Scope 3 emissions is challenging due to the complexity of supply chains and the absence of standardized, detailed shipment data. Traditional manual methods for estimating emissions are labor-intensive, inconsistent, and lack the accuracy needed for reliable environmental reporting. Companies are under growing pressure from regulators, investors, and consumers to disclose their carbon footprints transparently and accurately, especially concerning indirect emissions, which comprise a significant portion of their overall environmental impact.
The Project Scope
The project scope included:
- Designing an ML classifier capable of processing diverse shipment data inputs including logistics routes, transport modes, cargo details, and operational characteristics;
- Providing accurate estimation models for Scope 3 carbon emissions derived from classified shipment activities;
- Integrating the classification system with existing sustainability reporting and carbon management platforms;
- Ensuring scalability to accommodate extensive data volumes from global supply chains;
- Incorporating explainability features to allow users to understand the basis of emissions estimates and support regulatory audits.
The Solution
Sirma developed a sophisticated ML system utilizing supervised learning on rich datasets combining shipment records and emission factors. The model classifies shipments by categories such as transport type (air, sea, road), weight, and distance to feed into carbon emission calculation algorithms aligned with international standards like the Greenhouse Gas Protocol. To achieve accurate Scope 3 carbon emissions estimates, Sirma preprocesses shipment data through a comprehensive, multi-step approach designed to cleanse, consolidate, and contextualize diverse logistics information. Advanced data preprocessing and feature engineering improve model accuracy, while continuous training on new data ensures adaptability to evolving logistics patterns. The solution streamlines emissions calculation, delivering timely reports and actionable insights to sustainability managers.
Results
- Achieved highly granular and accurate classification of shipment activities, leading to precise carbon emissions estimation;
- Automated the traditionally manual and error-prone process, reducing time and operational costs;
- Supported compliance with global sustainability regulations and reporting standards;
- Empowered businesses with enhanced visibility into emissions hotspots, enabling targeted reduction strategies;
- Improved transparency and confidence for stakeholders, including regulators, investors, and customers.
Technologies
The solution capitalized on a modern AI and data stack, including:
- Machine learning libraries for model development and training;
- Big data tools to preprocess and manage extensive shipment datasets;
- Explainable AI frameworks to interpret model decisions and ensure transparency;
- APIs connecting emissions data with external sustainability and reporting systems.
Sirma’s Relationship with Client
Sirma specializes in using machine learning (ML) to tackle complex environmental data challenges, providing accurate and automated carbon emissions classification that supports sustainable business practices. The company builds long-term partnerships focused on collaboration and innovation, working closely with clients throughout their sustainability transformation journeys. By aligning AI solutions with the specific environmental goals and regulatory requirements of each client, Sirma delivers tailored and scalable technologies.
