Neural networks have received increasing attention recently, which provide a relatively effective and simple method for dealing with highly complex problems. In this paper, a neural network-based prediction model for consumer purchase decision is constructed. The quantile regression function can reveal the characteristics of the entire conditional distribution of the response variable, and then using the neural network structure, the nonlinear structure in the factors affecting consumer purchasing can be simulated. The article selects the data related to the sales of notebook computers from January to June 2024 as the research object for empirical research, and the results show that, after comparing and analyzing with the linear quantile regression model prediction method, it is clear that the model of this paper investigates and predicts with a higher degree of accuracy, and with a better goodness of fit. The five quartiles of 0.1, 0.3, 0.5, 0.7 and 0.9 were selected for prediction, word of mouth and quality service, which can promote consumer purchase decision. Higher selling prices date lower consumer purchase decisions. Notebook higher memory does not significantly promote consumer purchases and should be in line with the normal needs of consumers.
