The allocation of public funds for research is often based on assessments of universities’ performance, scientific excellence and societal impact. To improve the global competitiveness of selected top universities, national governments launch academic excellence initiatives (programs and projects) designed for them. Higher education institutions conduct competitions for high-quality research funded by these programs. The competitions are usually decided by expert reviews, which assess the research environment, expected research results, and the non-scientific impact of these results. Experts attempt to obtain a responsible allocation of limited public funds for research, but they lack the support of tools for making objective decisions and are often accused of bias. This paper presents the application of Support Vector Machines (SVM) in evaluating the scientific excellence of higher education research project proposals. This method allows for the analysis of large sets of diverse data and is the basis for reducing the subjectivity of assessments obtained only through peer reviews. The theoretical considerations presented are complemented by the results of empirical studies, which confirmed the usefulness of the proposed method and the possibility of relatively simple calculation of computational errors. The results obtained from theoretical considerations and empirical research are the basis for further research on improving the proposed method.