Leakage-Aware Bandgap Prediction on the JARVIS-DFT Dataset: A Phase-Wise Feature Analysis

TL;DR

This paper introduces a leakage-aware approach to bandgap prediction using a curated subset of the JARVIS-DFT dataset, demonstrating that physical descriptors and phase-wise modeling achieve high accuracy without leakage of band-structure information.

Contribution

It presents a systematic feature analysis and a three-phase modeling framework that controls for leakage, providing a new curated dataset and baseline for future research.

Findings

Tree-based models achieve R2 of 0.88-0.90 across phases.

Expanding descriptors does not significantly improve accuracy when leakage is controlled.

Dielectric tensor components are identified as key contributors.

Abstract

In this study, we perform a systematic analysis of the JARVIS-DFT bandgap dataset and identify and remove descriptors that may inadvertently encode band-structure information, such as effective masses. This process yields a curated, leakage-controlled subset of 2280 materials. Using this dataset, a three-phase modeling framework is implemented that incrementally incorporates basic physical descriptors, engineered features, and compositional attributes. The results show that tree-based models achieve R2 values of approximately 0.88 to 0.90 across all phases, indicating that expanding the descriptor space does not substantially improve predictive accuracy when leakage is controlled. SHAP analysis consistently identifies the dielectric tensor components as the dominant contributors. This work provides a curated dataset and baseline performance metrics for future leakage-aware bandgap prediction studies.