A Decision Tree Model for Inset-Fed Microstrip Patch Antenna

This work proposes a novel approach to modeling inset-fed microstrip patch antennas using Decision Tree (DT) algorithm. The technique involves a well-structured and systematic adjustment of the decision tree (DT) model's hyperparameters. This process includes evaluating the influence of key factors such as the number of trees, the depth of splits, and the learning rate on the overall performance of the model, we identify the most suitable configurations for accurately predicting antenna resonant frequencies. Furthermore, we investigate the relationship between resonant frequencies and various physical and geometrical factors, including material properties, patch dimensions, and feed location. To validate the effectiveness of our model, we evaluate the model's performance in comparison with several artificial intelligence techniques, such as Radial Basis Function (RBF) neural networks, Multilayer Perceptron (MLP) neural networks, and Gaussian Process Regression (GPR). The findings demonstrate that the proposed DT model outperforms these alternatives regarding metrics such as mean square error and coefficient of determination. Moreover, the outcomes align closely with those obtained using HFSS software, confirming the accuracy and reliability of our model for predicting resonant frequencies within the 1-10 GHz range.