Research on Consumption Trend Prediction and Market Regulation in the Network Economy Based on Deep Learning

As the online economy develops more rapidly, consumer behavior is becoming more complex and unpredictable, and traditional methods to predict consumer trends are under great challenge. Therefore, in the study, an enhanced version of the deep learning LSTM (Long Short-term Memory Network) model is proposed for the accurate prediction of consumption trends and the optimization of the network economy market regulation strategies. Unlike traditional LSTM model, an adaptive attention mechanism and a temporal feature enhancement module are implemented to the model. The adaptive attention mechanism can dynamically assign different weights for the input data, so as to give the model greater attention to important features. Time Series Feature Enhancement Module: The time series feature enhancement module further enhances the learning ability of the LSTM model by introducing other external market factors as factors during consideration, enabling the model to capture more complex consumer trends. Based on the experimental results, it is concluded that compared with the traditional LSTM model, proposed model has better performance in prediction accuracy and consumption trend stability.