inputoutputmodeller

inputoutputmodeller is a software framework for constructing and analyzing input-output models, focusing on the relationships between inputs and outputs within a system. It is designed for researchers and practitioners who need to quantify indirect effects, simulate changes in demand or resources, and compare alternative configurations in a transparent, reproducible way.

The framework provides data structures to represent nodes (sectors, processes, or components) and flows between them, as well as tools to assemble input-output matrices and other model components. It supports both static, linear representations such as Leontief-type models and more dynamic formulations that track changes over time. The package includes routines for solving systems of linear equations, computing Leontief inverses, and performing scenario analysis.

Users can calibrate coefficients from observed data, define final demand or policy shocks, and run simulations to observe how changes propagate through the network. The tool offers basic visualization capabilities, such as matrix views and flow diagrams, and can export results to common formats for reporting.

Data import is designed to accept common tabular formats (CSV, Excel, JSON) and to integrate with other data-processing workflows. The project emphasizes interoperability, providing interfaces or bindings for multiple programming environments, so models can be embedded in larger data science or decision-support pipelines.

Common use cases include economic sector analysis, supply chain risk assessment, energy-system modelling, and environmental impact studies where indirect effects matter. Typical workflows involve defining entities and connections, specifying demand or stress scenarios, running the model, and interpreting both direct and indirect outcomes.

As a software project, inputoutputmodeller is maintained as an open-source library with ongoing development. Users should be aware of assumptions inherent in input-output modelling, such as fixed coefficients and linear responses, and should validate results against empirical data and domain knowledge.