Selection of features is the mechanism that recognizes the most appropriate attributes and elimination of the redundant and insignificant attributes. These research focuses on a feature selection approach conducted using wrapper methods to predict the individual status/test outcome of the Ethiopian Demographic and Health Survey (EDHS-HIV/AIDS) data set for HIV / AIDS. The study uses three widely employed wrapper-based methods of feature selection to validate the efficacy of the proposed methods namely: Forward Feature Selection (FFS), Backward Feature selection (BFS) and Recursive Feature selection (RFS). We used seven classification algorithms for the purpose of testing selected feature performance, and each classifier output is evaluated using accuracy, precision, recall, f1-score, and ROC. Among the algorithms, the classifiers namely Random Forest, K-Nearest neighbors and Gradient Boosting classifiers achieve higher accuracy levels on the EDHS-HIV/AIDS dataset than others after wrapper method applied. In our research, we have proved that the importance of specified feature selection methods is improving the learning algorithm performance.