This paper proposes a machine learning based model predictive controller for voltage and frequency regulation in islanded microgrid. The conventional PI controller, Finite Set Controller based Model Predictive Controller (FCS-MPC), FCS-MPC combined RF (FCS-MPC-RF) approaches are considered. The primary goal is to achieve reliable power management while ensuring stable voltage and frequency during load disturbances. Simulation results shows that conventional controller exhibits noticable deviations in frequency and voltage. While FCS-MPC, FCS-MPC-RF controller exhibits better performance providing stable voltage and frequency. The RMSE performance metric is considered to evaluate the performance of predicted model. Overall, combining Random Forest with MPC improves the performance of the system and best solution for microgrid applications.