Intelligent Accounting Cost Fusion Analysis and Prediction Model Based on Multi-source Heterogeneous Data

The purpose of this study is to explore the intelligent accounting cost fusion analysis and prediction model based on multi-source heterogeneous data, so as to solve the challenges of traditional accounting cost analysis methods in dealing with complex business environment. By integrating multi-source data such as enterprise internal system, external data and public data, this study constructs a comprehensive data set, which provides sufficient support for the establishment and training of the model. In the model establishment stage, deep learning models were adopted in the study, especially models combining Convolutional Neural Networks (CNN) and Long Short Term Memory Networks (LSTM), to model and predict cost data. The empirical analysis results show that our intelligent accounting cost fusion analysis and prediction model shows good results in predicting costs, and it is found that there is a certain correlation between costs and market share of enterprises. This research provides important cost management and marketing decisions reference for enterprises, which has important theoretical and practical significance.