Random Forest Regression for Predicting Metamaterial Antenna Parameters
Nazmia Kurniawati, Dianing Novita Nurmala Putri, Yuli Kurnia Ningsih
Abstract
Metamaterial is an artificial substance that has unique properties such as negative refractive index and negative permittivity that do not exist naturally in the universe. Metamaterial has been extensively used in antenna applications because of its numerous advantages. In antenna applications, the Split Ring Resonator (SRR) structure in the metamaterial antenna can improve antenna performance. In this paper, we use random forest regression which is part of machine learning algorithm to predict antenna parameters, such as gain, Voltage Standing Wave Ratio (VSWR), bandwidth, and return loss. Based on prediction result, number of estimator that resulted in lowest MAE for gain is 3 while for MSE is 2. For VSWR the lowest MAE and MSE is reached when the number of estimator is 8. For bandwidth, lowest MAE is achieved when the number of estimator is 1 while for MSE is 8. Return loss reaches the lowest MAE and MSE when the number of estimator is 24.