A machine learning-based aging gene feature selection and classification experiment is designed as the experimental content of the “Machine Learning Basics” course for intelligent medical engineering and other majors. In this experiment, the data set is obtained by mapping aging genes to gene ontology, feature selection methods are used to deal with feature redundancy in gene ontology, and classification models such as naive Bayesian and support vector machines are used to classify aging genes. The experiment is implemented with Python language and Scikit-learn framework. In addition to the built-in methods of the framework, a hierarchical feature selection method based on the statistical properties of the data and the uniqueness of the test sample is designed to eliminate the hierarchical redundancy among features. Experimental results show that effective feature selection methods can significantly improve the results of aging gene classification.