Fresh agricultural commodities have been entering the era of network marketing. However, the coverage population is still relatively small. In this paper, more than 400 online shopping customer survey data are statistically analyzed based on perceived risk multidimensional model by factor analysis method to classify the potential customers' perceived risk, concluding that food safety risks, mental health risk, relative convenience risk, liquidity risk, privacy risk and time risk are the most important risk factors that impact potential customers online shopping of fresh agricultural commodities. By using customers prediction model which is based on the classification and prediction methods to mining potential customers, it comes to the conclusion that men are more likely to purchase fresh agricultural commodities online, specifically, in the male sample, those whose average monthly net purchase cost equals to or is higher than 51 yuan or whose online shopping time equals to or is longer than 3 years and at the same time whose age is younger than 30 are the most potential customers. Finally, it puts forward corresponding countermeasures and suggestions from the perspectives of risk control and effective customer acquisition.