Abstract
This paper explores the use of Decision Tree algorithm in the development of small signal model of GaN HEMT. In this stage, each measured s-parameters are modelled separately exploiting the bias, frequency and geometry dependence of the device as input predictors. This necessitates the tuning of parameters using Bayesian optimization and Random search algorithms. The outcome in terms of MSE and MAE demonstrates that the Random search algorithm gives a superior agreement with the measured values for the entire frequency range. Subsequently, the developed model is incorporated in the commercial CAD environment and a class-F power amplifier is designed to highlight the seamless integration ability and effectiveness of the developed model.