Documentation Index
Fetch the complete documentation index at: https://docs.mangrovesystems.com/llms.txt
Use this file to discover all available pages before exploring further.
What you’ll learn in this lesson:
- Review the complete end-to-end data flow of a production-ready project
- Build a configuration checklist for model completeness
- Understand testing strategy: unit, integration, and end-to-end testing
- Identify the key verification points in the data pipeline
The complete Mangrove Biochar data flow
Here is the full end-to-end pipeline for your training project:The data flow goes in one direction: events → models → batches → ledgers → reports. Each layer depends on the one before it. A missing link at any layer will cause downstream failures.
Configuration checklist
Before testing, walk through this checklist to verify every component is properly configured:Event types and datapoints
- All required event types are created in Data Inputs > Input Settings
- Each event type has the correct datapoints with appropriate value types
- Datapoint slugs match what the models expect
- Required vs. optional flags are set correctly
Static inputs and constants
- Emission factors are stored as static inputs in the resource library
- Conversion constants (e.g., 3.67) are defined in the model
- Values and units are correct
Models
- All model nodes reference the correct datapoint slugs
- Operators are appropriate for each calculation step
- Intermediate nodes produce expected values in the Console
- Output datapoint types match what the batch type expects
- Model version is saved and validated
Batch types and partitioning
- Batch type is configured with the correct partition cadence
- Output datapoint types in the batch type match the model outputs
- Feedstock partitioning is configured if needed
Ledgers and allocations
- Ledgers are configured for each stage (if using mass balance )
- Batch types are linked to the correct ledgers
- Allocation rules between ledgers are set up
Testing strategy
Testing a production accounting project has three levels:- Unit testing
- Integration testing
- End-to-end testing
Test individual components in isolation.What to test:
- Each model node produces the correct output for known inputs
- Each event type accepts valid data and rejects invalid data
- Each emission factor produces the expected result
Key verification points
At each layer of the pipeline, there are specific things to verify:| Layer | What to Check | Common Failure Mode |
|---|---|---|
| Events | All required datapoints present, valid values | Missing datapoints, wrong units |
| Models | Intermediate values match expectations | Wrong slug reference, missing division by 100 |
| Batches | Correct number generated, outputs match | Wrong partition cadence, mismatched output slugs |
| Ledgers | Balance reconciles (in = out + change) | Missing allocation rules, wrong batch type linkage |
| Reports | Aggregated totals are correct | Incomplete batch coverage, wrong period selection |
Check your understanding
What are the 5 configuration layers in the complete data flow?
What are the 5 configuration layers in the complete data flow?
Events (data entry) → Models (calculations) → Batches (partitioned outputs) → Ledgers (mass balance tracking) → Reports (aggregated for verification). Each layer depends on the previous one being correctly configured.
What is the difference between unit testing and integration testing in this context?
What is the difference between unit testing and integration testing in this context?
Unit testing verifies individual components in isolation — a single model node, a single event type, a single emission factor. Integration testing verifies the connections between components — that events feed correctly into models, models produce correct batch outputs, and allocation distributes properly. Unit tests catch formula errors; integration tests catch configuration mismatches.
Next, learn specific testing patterns and techniques in Lesson 5.2: Testing Patterns.