Related Experiment Video
Updated: Jul 9, 2025
11:38
Influence of Hybrid Perovskite Fabrication Methods on Film Formation, Electronic Structure, and Solar Cell Performance
Published on: February 27, 2017
18.5K
Predicting the bandgap and efficiency of perovskite solar cells using machine learning methods
Asad Khan1, Jeevan Kandel1, Hilal Tayara2
1Graduate School of Integrated Energy-AI, Jeonbuk National University, Jeonju, 54896, South Korea.
Molecular Informatics
|December 5, 2023
Summary
Machine learning models accurately predict perovskite solar cell bandgaps and efficiency. This approach, using CatBoostRegressor and SHAP analysis, offers a faster, cost-effective alternative to traditional methods for discovering new materials.
Area of Science:
- Materials Science
- Renewable Energy
- Computational Chemistry
Background:
- Accurate prediction of perovskite solar cell bandgaps and efficiency is vital for solar energy applications.
- Traditional experimental and theoretical methods are often time-consuming and expensive.
- Machine learning (ML) offers a computationally efficient alternative for predicting material properties.
Purpose of the Study:
- To develop and evaluate ML models for rapid and accurate prediction of perovskite solar cell bandgaps and efficiency.
- To identify the most effective ML model and understand feature contributions using SHAP analysis.
- To demonstrate the robustness and generalizability of the developed ML model.
Main Methods:
- Trained various ML models using a dataset of reported experimental data for perovskite solar cells.
- Selected CatBoostRegressor as the best-performing model for predicting bandgap and efficiency.
- Employed k-fold cross-validation for model evaluation and SHAP (Shapley Additive Explanations) for feature importance analysis.
- Validated the model's performance on an independent dataset.
Main Results:
- The CatBoostRegressor model demonstrated superior performance in approximating both bandgap and efficiency.
- SHAP analysis provided valuable insights into the contribution of different features to the model's predictions.
- The model showed robustness and generalizability when tested on an independent dataset, confirming its reliability beyond the training data.
Conclusions:
- ML-based approaches, particularly the CatBoostRegressor model combined with SHAP interpretation, provide a computationally efficient and accurate method for predicting perovskite solar cell properties.
- This ML approach can significantly accelerate the discovery of novel perovskite materials for solar cell applications.
- The developed model and its code are publicly available to the research community to foster further advancements.
