Africa_MATRIX_Trained_Models

This Figshare item provides the trained Random Forest (ranger) demographic models used in the Africa MATRIX forest dynamics framework.

The archive includes serialized R model objects for:

(i) mortality,

(ii) upgrowth (diameter increment), and

(iii) recruitment.

Each model was trained on harmonized African forest inventory data using the ranger package and includes the selected feature set and tuned hyperparameters. These models were used for cross-validation, forward interval validation (Inventory 1 → Inventory 2), and long-term forest projection analyses under constant and climate-change scenarios.

All preprocessing, training, validation, and simulation code used to generate these models is available at:

github.com

Processed training data and validation outputs are archived separately on Figshare.