rapsimng.decide

rapsimng.decide provides a high-level interface for analysing APSIM Next Generation simulation outputs through structured decision workflows.

The package coordinates domain-specific decision modules (e.g. cultivar suitability, experiment design) and returns standardised, reproducible decision reports.

Installation

Currently on Github only. Install with:

remotes::install_github('byzheng/rapsimng.decide.cultivar')

Overview

Crop modelling often involves multiple steps:

  1. Run APSIM simulations
  2. Analyse outputs for a specific decision
  3. Summarise results for reporting or interpretation

This package focuses on Step 2 and 3 by providing a unified entry point for decision analysis.

Role in the Ecosystem

The package sits within a layered design:

  • rapsimng
    Handles APSIM file manipulation and simulation

  • rapsimng.decide.*
    Implements domain-specific decision logic
    (e.g. cultivar, experiment design)

  • rapsimng.decide
    ✅ Coordinates decision packages
    ✅ Standardises outputs
    ✅ Provides a consistent user interface

  • agrillm
    Interprets user intent and explains results

Key Concept

This package does not define decisions.
It coordinates how decisions are evaluated and reported.

Usage

High-level entry point 

report <- evaluate_decision(
  data,
  decision = "cultivar",
  context = list(...),
  criteria = list(...)
)

Output Structure

All decisions return a consistent structure:

list(
  meta     = list(...),
  metrics  = list(...),
  tables   = list(...),
  figures  = list(...)
)

This structure is designed for:

  • reproducible workflows
  • direct use in Quarto (qmd)
  • downstream interpretation (e.g. agrillm)

Design Principles

1. Separation of Responsibilities

  • Decision logic lives in rapsimng.decide.*
  • This package only orchestrates and standardises outputs

2. Transparency

  • All assumptions and criteria are explicit
  • No hidden optimisation or recommendation

3. Reproducibility

  • Outputs are deterministic and structured
  • Suitable for reporting and publication

Scope

  • APSIM Next Generation outputs only
  • Input: data.frame / tibble
  • No dependency on APSIM runtime

Not in Scope

  • APSIM simulation setup or execution
  • Domain-specific decision logic
  • User intent interpretation
  • Automated decision-making

Extending the System

New decision types can be added by creating packages such as:

  • rapsimng.decide.cultivar
  • rapsimng.decide.experiment

Each package:

  • implements its own metrics, tables, and figures
  • returns a standard decision report object

rapsimng.decide will automatically integrate them through its interface.

Key Insight

This package provides a decision framework,
not decision answers.

Status

Early-stage infrastructure for:

  • modular decision analysis
  • scalable modelling workflows
  • integration with AI-assisted interfaces