The social capital theory is selected as the theoretical basis, and the decision-making of household commercial insurance purchase is taken as the research object. By using the frontier theory of big data, an improved K-means clustering algorithm is proposed to segment customers first, then the results of the segmentation are predicted and analyzed. Finally, an insurance consumer purchase behavior model is proposed, and an empirical test is carried out according to the proposed model. Social capital and its three dimensions can promote the possibility in family commercial insurance, thus ultimately promoting residents' families to make purchase decisions on commercial insurance.