This paper suggests the suitable network model in machine learning to estimate the switching power loss for the DC-DC converter. A boost converter topology is selected to compare the suitable network model based on their performance. Unlike traditional simulation approach, empirical formulae-based prediction approach is used to estimate the switching loss of power MOSFET. To obtain the data set, three variable parameters such as switching frequency, duty ratio, load resistance for fixed input voltage is considered. Four different network models namely Feed Forward Back Propagation, Cascaded Forward Back Propagation, Layer Recurrent and Elman Back Propagation network models are compared to obtain the accurate predicted value of switching loss of the converter.