A systematic and general machine learning approach to build a consistent data set from different experiments

Experimental data from different sources present challenges due to variability and noise from various experimental conditions, apparatuses, and environmental factors. In this work, we propose a general method to address these challenges to build a consistent data set employing different thermal conductivity experimental data sets of methane from the liquid, vapor, and supercritical phases. Methane is a key hydrocarbon with extensive industrial and environmental applications. The method is based on machine learning (ML) techniques, which are used to consistently integrate data from various experimental sources compiled by the National Institute of Standards and Technology (NIST) database. Different ML algorithms are used for this purpose. Our findings indicate that ML models trained on raw experimental data yield predictions closer to the NIST’s processed data than the original raw experimental data, thus demonstrating the models’ ability to generalize from heterogenous, noisy, and untreated data sets. The proposed ML approach is general and efficient in handling complex and heterogeneous data to deliver reliable predictions without extensive preprocessing.