In the real production and sales scenarios of agricultural products, the network structure formed by consumers co-purchase behavior is very complex and changeable. Although the community discovery algorithm can effectively dig out the hidden information behind the co-purchase behavior, there are problems that the analysis results are not easy to understand, and the supporting decision-making conditions are insufficient. Because of the widespread application of community discovery algorithms in the analysis of co-purchase networks, and the ability of visualization technology to present the analysis results, a visual analysis method for the co-purchase networks of agricultural products based on community discovery was proposed. Firstly, the method used the community discovery algorithm Clauset-Newman-Moore (CNM) to divide the agricultural product co-purchase network. Secondly, the quantity of agricultural products in different communities in the network structure, the frequency of co-purchase behaviors, and the proportion of the price mode of agricultural products were analyzed, and then interactive analysis on the information of customers who co-purchase a certain agricultural product in each community was conducted. Finally, the analysis results were displayed interactively and visually. According to the visual interface, some behavioral rules of co-purchases were obtained, and then their consumption rules were deeply explored. In order to better present the visual analysis method, a set of dynamic sales data of agricultural product in Qingdao area were interactively explored and analyzed through the design of a visual analysis interface, and the sales model found can not only inspire the improvement and optimization of the manufacturers marketing methods, but also can help consumers to better choose agricultural products that suit them.