Enterprise marketing data mining method based on decision tree algorithm

In the context of digital transformation, enterprises are faced with the challenge of insufficient value density of multi-source heterogeneous data, and it is urgent to improve the effectiveness of marketing decision-making through efficient data mining technology. Based on decision tree algorithm, this paper proposes an improved C5.0 model for enterprise marketing data mining. By introducing information entropy optimization feature selection mechanism, combining clustering algorithm and dynamic threshold partitioning strategy, the whole process analysis framework of "data preprocess-decision model-rule generation" is constructed. Based on UCI data set, the results show that the improved C5.0 algorithm maintains 86.7% classification accuracy, while the computational efficiency is improved by 14.5% compared with the traditional decision tree, and the response time is reduced to 1.87 seconds. Compared with BP neural network and ID3.0 algorithm, its comprehensive performance has significant advantages. The research proves that the decision tree framework of fusion cluster analysis can effectively solve the problems of high-dimensional noise and attribute redundancy in marketing data, provide technical support for accurate customer clustering and dynamic marketing strategy formulation, and have strong interpretable and real-time application value.