Machine learning (ML) is a powerful tool for data analysis, enabling organizations to extract insights from large and complex datasets. This study investigates the impact of various machine learning algorithms, missing value imputation techniques, and dimensionality reduction techniques on the efficiency of models that predict the Social Institutions and Gender Index (SIGI). A complete factorial design is employed to systematically examine the interactions between these factors. The SIGI dataset, characterized by high dimensionality and missing values, is used as a benchmark. Experimental results demonstrate that random forest, in conjunction with weighted KNN imputation, histogram gradient tree imputation, and Uniform Manifold Approximation and Projection (UMAP), achieves the highest accuracy of 91%. These findings provide valuable insights for researchers and practitioners seeking to apply machine learning to complex social science datasets.