Data-Driven Prediction of Composition–Processing–Microstructure Relationships for Fe-Intermetallic Phases in Recycled Al–Si–Fe–Mn Cast Alloys

This repository contains the machine-learning code used to predict the formation and morphology-resolved characteristics of Fe-rich intermetallic compounds in recycled Al–Si–Fe–Mn cast alloys.

It includes interpretable GAMI-Net classifiers for morphology formation probability, ordinal cumulative (threshold) models for microstructural descriptor prediction, and utilities for generating pairwise interaction heatmaps and composition–processing landscapes. The code supports reproducible training, evaluation, and visualization of morphology selection, interaction effects, and morphology-conditioned microstructural burden under sparse experimental data.