Segmentation in Structural Equation Modeling Using a Combination of Partial Least Squares and Modified Fuzzy Clustering

The application of a structural equation modeling (SEM) assumes that all data follow only one model. This assumption may be inaccurate in certain cases because individuals tend to differ their responses, and failure consider heterogeneity threaten the validity SEM results. study focuses on unobservable heterogeneity, where difference between two or more sets does not depend observable characteristics. In this study, we propose new method for estimating parameters containing unobserved within assume arises from outer model inner combines partial least squares (PLS) modified fuzzy clustering. Initially, each observation was randomly assigned weights selected segment. These continued iteratively updated using specific objective function. sum weighted residual resulting models PLS-SEM is an function must minimized. We then conducted simulation evaluate performance by considering various factors, including number segments, specifications, variance endogenous latent variables, indicators, population size, distribution variables. From its actual data, conclude proposed can classify observations into correct segments precisely predict