| dc.description.abstract | This research investigates the integration of Particle Swarm Optimization (PSO) with Structural
Equation Modeling (SEM) to enhance model identification and hyperparameter adjustment. The objective
function in the PSO-SEM algorithm combines various fit measures and considers the trade-off between
model complexity and fit. For this analysis two datasets that provide different model complexity are
employed. These datasets undergo similar preprocessing steps, including handling missing data and
partitioning into training and validation sets. The PSO-SEM algorithm is applied to optimize the model
fit, and the performance of the final models is evaluated using the validation sets. Through the exploration
of different hyperparameter combinations and values, valuable insights are obtained regarding their
relative importance and optimal settings. Additionally, the transferability of the selected hyperparameters
across different datasets is assessed, and further testing and refinement are conducted to ensure their
applicability in diverse contexts. The integration of PSO with SEM offers a flexible and efficient
approach for addressing complex problems and uncovering latent relationships in data, while the
computational time can be adjusted in each problem, by appropriately tuning parameters such as the
number of iterations while sacrificing a certain degree of the accuracy. Generally, this research
contributes to the fields of metaheuristics and structural equation modeling by exploring the integration of
PSO with SEM for enhanced model identification and hyperparameter adjustment. The findings offer
valuable insights and practical implications for researchers engaged in solving complex problems and
uncovering latent relationships in data. Furthermore, the results encourage additional investigations,
including different metaheuristic algorithms, a variety of datasets and the application of more
hyperparameter combinations. | |
