Pleurotus pulmonarius — EB Prediction

This repository contains R scripts and data for substrate characterization and biological efficiency (EB%) modeling in Pleurotus pulmonarius.

Contents

• 01_pipeline_EB.R — Main pipeline:
  • Random Forest (Initial vs Consumed composition)
  • Decision Trees (pruned ~8 leaves, CV10 OOF)
  • Bootstrap stability + heatmaps
  • Bland–Altman comparison with new data
• data/ — Example input data (if shared)
• outputs/ — Generated results (PNG, CSV, TXT)

Outputs (generated automatically)

Running 01_pipeline_EB.R will generate:

• RF_EB_INICIAL_metrics_split.csv | RF_EB_CONSUMO_metrics_split.csv
• RF_EB_INICIAL_metrics_cv10.csv | RF_EB_CONSUMO_metrics_cv10.csv
• RF_EB_INICIAL_importance_bootstrap.csv | RF_EB_CONSUMO_importance_bootstrap.csv
• RF_EB_INICIAL_heatmap_top2.png | RF_EB_CONSUMO_heatmap_top2.png
• DT_EB_INICIAL_8leaves.png | DT_EB_CONSUMO_8leaves.png
• DT_EB_INICIAL_rules.txt | DT_EB_CONSUMO_rules.txt
• Predicciones_EB_nuevos_datos.csv
• BA_RF_inicial.png | BA_RF_consumo.png | BA_DT_inicial.png | BA_DT_consumo.png
• BlandAltman_summary_nuevos_datos.csv

Requirements

• R >= 4.2
• Packages: ranger, caret, dplyr, ggplot2, viridisLite, readr, tidyr, rpart, rpart.plot, scales

Usage

Clone the repo and run the pipeline:

## How to run

1) Install dependencies and run the full pipeline:
```r
# from the repo root in R/RStudio
source("scripts/main.R")

source("01_pipeline_EB.R")