Glossary¶
This glossary defines core terms used throughout the Electric Barometer ecosystem. The definitions here are conceptual and system-level, rather than implementation-specific. Where formal or mathematical definitions exist, they are provided in the relevant technical notes or papers.
The purpose of this glossary is to establish a shared vocabulary and reduce ambiguity across evaluation, decisioning, and governance discussions.
Asymmetric Cost¶
A property of a decision problem in which different types or directions of error incur different consequences. In forecasting contexts, this often refers to situations where under-forecasting and over-forecasting have unequal operational or economic impact.
Decision Regret¶
The realized cost or consequence of a decision made under uncertainty, relative to an alternative decision that could have been made with perfect information. Decision regret focuses on outcomes rather than prediction error alone.
Decisioning¶
The process of selecting an action based on available information, evaluation results, and explicit policy. Decisioning is inherently normative and reflects organizational priorities, risk tolerance, and constraints.
Evaluation¶
The process of measuring and comparing the behavior of forecasting systems relative to explicit criteria. Evaluation produces structured, reproducible measurements intended to inform decisioning, not replace it.
Forecasting System¶
A combination of data inputs, feature transformations, models, and configurations that produces forecasts. In Electric Barometer, forecasting systems are evaluated and selected as whole systems rather than isolated models.
Governance¶
The set of rules, policies, and controls that determine how evaluation results are interpreted, how decisions are made, and how those decisions are reviewed, audited, or revised over time.
Metric¶
A quantitative measure used to summarize some aspect of forecasting system behavior. Metrics provide information for evaluation but do not, by themselves, encode decision policy.
Operational Context¶
The real-world environment in which forecasts are used, including constraints, costs, timing, and downstream consequences. Operational context determines how forecast errors translate into impact.
Policy¶
An explicit set of rules or preferences that govern how evaluation outputs are translated into decisions. Policies encode tradeoffs, priorities, and risk posture rather than technical performance alone.
Reproducibility¶
The property that the same inputs, evaluation criteria, and decision policies yield the same results and decisions when applied consistently. Reproducibility is a core requirement for auditability and governance.
Selection¶
The act of choosing one forecasting system or configuration from a set of evaluated alternatives, based on evaluation outputs and decision policy.
Sensitivity¶
The degree to which evaluation outcomes or decisions change in response to variations in assumptions, parameters, or inputs. Sensitivity analysis helps reveal tradeoffs and robustness.
Tie-breaking¶
A policy-driven mechanism for resolving situations in which multiple forecasting systems perform similarly under evaluation criteria. Tie-breaking rules prevent arbitrary or unstable decisions.
Tradeoff¶
A situation in which improving performance along one dimension degrades performance along another. Tradeoffs are unavoidable in multi-objective decision problems and must be managed explicitly.
Uncertainty¶
The inherent lack of certainty in future outcomes, forecasts, or system behavior. Uncertainty motivates the separation of evaluation from decisioning and the use of policy-driven choices.
Workflow¶
An ordered sequence of stages through which data, forecasts, evaluations, and decisions flow. In Electric Barometer, workflows are designed to be explicit, reproducible, and inspectable.
Readiness¶
A measure of how prepared a forecasting system or decision policy is to support operational action, considering risk, uncertainty, and cost. Readiness emphasizes suitability for use rather than raw performance.