The classification decision tree fused with neural network realizes the intelligent transformation of data
Qi Wang, Xiaomin Zhu, Xinming Wang, Min Zou, Jinan Guo, Yalin Wang · 2022
The data preprocessing stage is a crucial step in the realization of data analysis and the establishment of algorithm models. However, the current data preprocessing including data cleaning, data transformation and other processing methods are not perfect, so a lot of time and human capital are often spent in data cleaning. In order to adapt to the development of artificial intelligence, it is necessary to get rid of traditional processing technologies and tools, and propose methods that can process data through inherent models such as artificial intelligence. This paper takes the transaction amount data in the bidding data as an example to solve the problem of non-uniform units and forms of the transaction amount data, and realize the standardized and unified intelligent processing process of the data. This paper proposes a classification decision tree model (Decision Tree fused with Neural Network, DTFNN) fused with neural network. By training a BP neural network with recognition function for specific characters, it is integrated into the classification decision tree, so as to improve the decision tree. The feature level of classification prevents the occurrence of over-fitting and under-fitting, and realizes the purpose of automatic data classification; then the classified data is transferred to the ICS (Input-Compute-Select) which includes three stages of input, calculation and selection. In the calculation model, a weight matrix of real-time conversion rate is constructed, so as to realize the intelligent conversion of data and simplify the process of data processing.