A model for the classification of 70 HIV-1 protease crystal structure binding pockets to one of its complexed FDA approved protease inhibitors utilizing a hybrid data mining modeling approach has been developed. 456 chemical descriptors have been derived from the binding pocket structure of each crystal structure. Two hybrid approaches were developed, Random Forest-Linear Discriminant Analysis (RF-LDA) and Random Forest-Logistic Regression (RF-LR). Random Forest is used as a feature selection proxy, selecting for the most relevant descriptors used to train its classification model. The top ranked descriptors are then used to train the subsequent LDA and LR models. The classification performance of LDA and LR are compared against the Random Forest classifier used to perform the feature selection. As a model validation step, hierarchical clustering of the top ranked descriptors is performed to verify the descriptor selection by Random Forest can group together the binding pocket structures based on their complexed ligands. Analysis of the top ranked chemical descriptors would play a crucial role in understanding the HIV-1 protease binding pocket in terms of its drug resistance and protein-ligand interactions.