Feature Selection in Machine Learning for Perovskite Materials Design and Discovery.

Area of Science:

• Materials Science
• Computational Materials Science
• Chemical Informatics

Background:

• Perovskite materials are vital in materials science due to their exceptional photoelectric properties and complex structures.
• Machine learning (ML) is increasingly integral to perovskite material design and discovery.
• Feature selection is a critical dimensionality reduction technique within the ML workflow for materials science.

Purpose of the Study:

• To review recent advancements in applying feature selection to perovskite materials.
• To analyze publication trends in ML for perovskite research.
• To summarize the ML workflow in materials science and introduce common feature selection methods.

Main Methods:

• Analysis of publication trends related to ML in perovskite materials.
• Summarization of the typical ML workflow for materials design.
• Introduction to various feature selection techniques.
• Review of feature selection applications across different perovskite types (inorganic, hybrid organic-inorganic, double perovskites).

Main Results:

• The study analyzes the growing trend of ML applications in perovskite research.
• It details the ML workflow and common feature selection methods.
• Applications of feature selection are reviewed for inorganic perovskites, hybrid organic-inorganic perovskites (HOIPs), and double perovskites (DPs).

Conclusions:

• Feature selection plays a pivotal role in optimizing ML models for perovskite material design.
• The review provides insights into current applications and identifies future research directions.
• Advancements in feature selection are essential for accelerating the discovery of new perovskite materials with desired properties.