The goal of this research is to optimize a static and dynamic compact thermal model for a ball grid array (BGA) package using experimental data. The general objectives of thermal modeling are to increase the accuracy of electrical analysis to enhance the performance of the electronic systems. The project is focused on generating the static and dynamic compact thermal model of a Bipolar Junction Transistor (BJT) and a Ball Grid Array (BGA) based on experimental results of infrared (IR) camera system , so that the steady state and transient thermal behaviors of the package could be predicted fast with required accuracy. The approach proposed by a previous study based on generation of dynamic compact thermal model of a BGA package using simulation tools, is extended in this work to generate the static and dynamic compact model of the same package represented by a RC (thermal) network or admittance matrix based upon a methodology which couples different layers of experimental data to the error minimization notion of the problem. This optimization problem sets the temperature profile experimental data as a standard and compares the compact model's computed temperature and refines itself with a feedback, until reaching a desired point.