Statistical approach for automated weighting of datasets: Application to heat capacity data

An essential step in calculation of phase diagrams (CALPHAD) assessments is assigning relative weights to different datasets.Nonetheless, there is no consensus as to the best approach regarding this issue. Currently, such an assignment of weights for experimental or first-principles data is performed manually based on the knowledge and experience of the researcher. Since the existing manual treatment is tedious and subjective, manipulation of such data is rapidly advancing towards automated procedures through statistical and data mining tools. In this study, we propose an automated approach to determine the weight of datasets based on the K-Fold Cross-Validation method, modified under the conditions that each fold is selected non-randomly and contains an unequal number of observations. Applying this approach enables researchers to evaluate the reliability of each dataset involved in an assessment and quantify the impact of weighting by statistical analysis of the corresponding model. We demonstrate theefficacy of this method through the evaluation of heat capacity data of pure iron and magnesium.Keywords: Weighting, K-fold Cross-Validation, Heat capacity, CALPHAD