Multidimensional analysis and prediction based on convolutional neural network

At present, artificial intelligence methods such as machine learning are widely used in E-commerce enterprises, but the disconnection between business practice and prediction technology is still a real challenge for E-commerce enterprises. Firstly, this paper focuses on the actual business of E-commerce enterprises, carries on a multi-dimensional analysis of the influencing factors of E-commerce sales, refines various factors affecting E-commerce sales, further summarizes the feature construction work of E-commerce sales prediction, constructs the feature project of sales prediction, and provides reference for the practical application of E-commerce enterprises. Secondly, an E-commerce sales forecasting model based on Convolutional Neural Network (CNN) and soft computing is proposed. The model adopts the feature learning of CNN's AlexNet and integrates the attention mechanism. Finally, based on the data of E-commerce enterprises, this paper compares the prediction effects of other conventional machine learning models. The experimental results show that the CNN based fusion prediction model proposed in this paper can improve the accuracy rate, have better prediction performance, and provide an effective in-depth learning method.