Ecosystem Map¶

The Electric Barometer ecosystem is composed of a set of focused repositories, each responsible for a specific layer of the forecasting, evaluation, and decisioning lifecycle. These components are designed to work together through shared concepts and contracts, while remaining independently versioned and governed.

This page provides a high-level map of the ecosystem and explains how the pieces fit together.

Design philosophy¶

The Electric Barometer ecosystem follows a small number of design principles:

Separation of concerns — forecasting, evaluation, policy, and decisioning are distinct layers
Explicit contracts — assumptions and interfaces are made visible
Independent evolution — repositories are versioned and released independently
Centralized concepts, distributed implementation — core ideas live in this documentation; implementation details live in their respective repositories

The goal is clarity, not consolidation.

Ecosystem at a glance¶

At a high level, the ecosystem spans five functional layers:

Concepts and Documentation Define the mental model, vocabulary, and system structure.
Features and Adapters Prepare inputs and connect external systems.
Metrics and Evaluation Measure forecasting system behavior under explicit assumptions.
Optimization and Policy Resolve tradeoffs and translate evaluation into decisions.
Research and Formalization Provide mathematical and theoretical grounding.

Each layer is represented by one or more repositories.

Core documentation¶

eb-docs¶

This repository serves as the central documentation hub for the ecosystem.

It contains:

Conceptual foundations
Guides and workflows
Ecosystem-level references
Navigation and integration across repositories

It does not duplicate code-level documentation. Instead, it routes readers to the appropriate source of truth.

Features and data preparation¶

eb-features¶

The features repository defines reusable feature transforms used by forecasting systems.

Responsibilities include:

Feature transform definitions
Input/output contracts
Validation logic
Feature-level documentation

Feature transforms are treated as governed, reusable components rather than ad hoc preprocessing steps.

See Add a Feature Transform for contribution guidance.

Adapters and integration¶

eb-adapters¶

The adapters repository contains integration layers that connect Electric Barometer to external systems and environments.

Adapters handle:

Data ingestion and extraction
Schema translation
Environment-specific I/O
External service integration

Adapters isolate operational concerns from evaluation and decision logic.

See Add an Adapter for details.

Metrics and evaluation¶

eb-metrics¶

The metrics repository defines evaluation measures used to assess forecasting system behavior.

It includes:

Cost-aware and decision-facing metrics
Parameterized metric definitions
Validation and interpretation guidance

Metrics measure behavior; they do not make decisions.

The flagship metric in this repository is Cost-Weighted Service Loss (CWSL). See Metrics for an overview.

Optimization and policy¶

eb-optimization¶

The optimization repository contains logic and frameworks for translating evaluation outputs into governed decisions.

It covers:

Cost ratio configuration
Policy definition and tuning
Tie-breaking and selection logic
Readiness-aware optimization

Optimization operates above metrics and models, encoding preference and intent explicitly.

See Optimization to understand this layer.

Forecasting and execution¶

eb-integration¶

The integration repository coordinates end-to-end execution across features, forecasting systems, evaluation, and optimization.

It provides:

Workflow orchestration
System assembly
Execution configuration
End-to-end validation

This repository ties the ecosystem together without redefining core logic.

Research and formal foundations¶

eb-papers¶

The papers repository contains formal technical notes and research papers that define the mathematical and theoretical foundations of the ecosystem.

It includes:

Formal metric definitions
Theoretical frameworks
Extended analysis and proofs

These artifacts serve as stable, citable references.

How the pieces fit together¶

A typical lifecycle flows as follows:

Feature transforms prepare inputs
Adapters connect data and execution environments
Forecasting systems generate predictions
Metrics evaluate system behavior
Optimization applies policy and readiness
Decisions are selected and governed
Releases preserve reproducibility

Each step corresponds to one or more repositories, coordinated through shared concepts rather than tight coupling.

Navigating the ecosystem¶

If you are:

New to Electric Barometer — start with Concepts
Running evaluations — see Guides
Working with metrics — explore Metrics
Defining policies — review Optimization
Looking for formal definitions — consult Papers

This documentation site provides the map; each repository provides the detail.

Where to go next¶

Continue with Start Here for onboarding
Read Concepts to understand system philosophy
Explore Guides for hands-on workflows

The Electric Barometer ecosystem is designed to scale in complexity without losing clarity. This map is your guide to how the pieces work together.