An Improvement of Parameter Estimation Accuracy of Structural Equation Modeling using Hybridization of Artificial Neural Network in the Entrepreneurship Structural Model

Abstract

In developing optimal entrepreneurship, several variables such as motivation, knowledge, intensity, and capacity are required to determine their relationship using the Partial Least Square-Structural Equation Modeling (PLS-SEM). The results show that entrepreneurial motivation and knowledge significantly affect intensity. Also, motivation and intensity significantly influenced capacity. The parameter estimator of PLS-SEM can be improved by applying hybridization to the Artificial Neural Network (PLS-ANN) using the 2:32:8:1 architecture in which motivation and intensity were the input while capacity was the output. The comparison parameter accuracy model measured by MSE, RMSE, and MAE shows the improvement accuracy by PLS-ANN better than PLS-SEM.