Methodological evaluation demonstrates execution of PLSpredict in structural equation modeling, highlighting standardized procedures for testing out-of-sample predictive power.
Key Points
To establish systematic guidelines and best practices for evaluating out-of-sample predictive performance in partial least squares structural equation modeling using the PLSpredict procedure.
Conceptual advancement of the PLSpredict algorithm using holdout sample procedures to generate case-level predictions at item and construct levels.
Demonstration of practical implementation and interpretation using an empirical tourism marketing dataset.
PLSpredict provides a standardized holdout sample framework to quantify out-of-sample predictive capabilities beyond traditional in-sample explanatory metrics.
Routine prediction-oriented assessments resolve discrepancies between explanatory fit and predictive utility in structural equation models.