Figure 2 | Scientific Reports

Figure 2

From: Machine learning with persistent homology and chemical word embeddings improves prediction accuracy and interpretability in metal-organic frameworks

Figure 2

Model performances for hMOF dataset and CO2 adsorption. Comparison of root-mean-square deviation (left), coefficient of determination (right) in predicting gas uptakes in CO2 for different features at different pressures for the hMOF dataset. The RMSD is low at lower pressures because the distribution of carbon dioxide adsorption capacity has low variance in this regime. The topological features consistently outperform the standard structural features at all pressures. The T + WE and T + S + WE models achieve the best performance in general.

Back to article page