To improve the classification quality of using the regulation rule in a zeroinflated binary (ZIB) model, the differential evolution (DE) and particle swarm optimization (PSO) algorithms are used in this study for optimization. The performance of the two algorithms is compared with the maximum likelihood estimation method. The elastic net regularization rule (ENR) is used to construct the loss function for the ZIB model, named the ENR-ZIB model, to prevent overfitting. The estimates of the model parameters of the ENR-ZIB model are obtained to minimize a specified loss function. Moreover, the classification performance of the obtained model is studied. Monte Carlo simulations are conducted to compare the performance of the ENR-ZIB model using two proposed optimization procedures with the maximum likelihood estimation method. Simulation results show that the proposed optimization procedures can have a good classification quality for the ENR-ZIB model.
