Data mining for methane adsorption over metal organic frameworks in literature

Abstract

A database containing 3000 data points for both gravimetric and volumetric CH4 storage/delivery in metal organic frameworks (MOFs) was analysed using machine learning tools via RStudio functions to extract knowledge for generalization. First, the database was passed through the reprocessing stage to deal with the missing values and inappropriate input variables. The cross-correlation analysis was also performed during this stage as well and the correlated variables were determined. Then the optimization process took place to find out the optimum conditions of the best models. First, the database was reviewed to observe the basic trends and patterns. It was then analysed using decision trees and artificial neural networks (ANN) to extract hidden information and develop rules and heuristics for the future studies. Five-fold cross validations were used in each analysis to test the validity of the models with data not seen before. Decision tree analyses were carried out using six user defined descriptors and two structural properties, separately. The crystal structure and the total degree of unsaturation were found to be the effective user defined descriptors, whereas the pore volume and maximum pore diameter, as structural properties, were sufficient to determine the MOFs having high CH4 storage/delivery capacity. Moreover, a high pore volume is always required as expected. In ANN analyses, models were performed by using the user defined descriptors and the structural properties separately. It was observed that the user defined descriptors were not sufficient to describe the CH4 storage/delivery capacity of MOFs, whereas the structural properties, especially pore volume, provided accurate CH4 storage/delivery prediction with low root mean square error (RMSE) and high coefficient of determination (R2) values.