The interpretability of black box machine learning models has been constantly evolving over the years, giving more assurance and confidence for them to be applied in crucial sectors of both companies and society. That being said, the use of these methods to receive insights on optimization models are not as popular and often result on these models being less explainable or interpretable. Motivated by this, this work seeks out to develop a machine learning equivalent of an optimization model named OptGen, whose goal is to obtain an optimal generation and transmission expansion plan, and with this model apply interpretability methods to analyze and understand its decision making process. The dataset to develop the black box model was generated utilizing 10,000 rounds of the OptGen model using Latin Hypercube sampling method, and with this dataset a neural network was trained on a supervised manner having its accuracy analyzed using statistical and graphical methodology. Lastly, utilizing a Shapley Values based method of interpretability, the behavior of the model was analyzed to gain insights on its decision making process. Based on the study done in this work, a strong correlation between battery and solar power insertion was observed by the model, as well as a pattern between limitations given to wind and hydro plants that are different than the observed for thermal, solar and battery.