Table 1 Predicting MOFs with high gas adsorption

From: Chemically intuited, large-scale screening of MOFs by machine learning techniques

CO2 Dataset H2 Dataset
Cutoff threshold t Correct predictions Random guessing Cutoff threshold t Correct predictions Random guessing
(gr_CO2/gr_MOF)    (gr_H2/gr_MOF)   
16 85.7% 21.6% 1.6 80.4% 45.9%
18 66.8% 18.6% 1.8 77.7% 32.7%
20 46.7% 13.7% 2.0 46.7% 23.5%
  1. Cutoff thresholds were used to classify MOFs into ones with high or low gas adsorption ability. Correct predictions using machine learning models correspond to correctly predicted values at least as large as t