In the process of defining automobile products, how to analyze consumers’ equipment needs is a big problem. The traditional method is mainly research method, which quantifies consumers' needs through different population surveys. However, the traditional quantitative method mainly relies on simple undetermined coefficient method and KANO model method, with relatively single dimensions and relatively high requirements for practitioners’ experience. With the help of machine learning method, more dimensions of data can be introduced and the analysis process is easier to visualize, so as to realize the migration and reuse of analysis model, and then support the digital management of products. With the help of more mature K-means clustering method, this paper uses the original data of consumer research to cluster, quantitatively analyze the equipment needs, and divide five clusters: prospective equipment, secondary equipment, essential equipment, charm equipment and desire equipment, so as to excavate the real needs of consumers and support the decision-making work of digital automobile equipment.