Affiliations: Department of Basic Subjects, Henan Institute of Technology, Xinxiang, China
Correspondence:
[*]
Corresponding author. Yutang Liu, Department of Basic Subjects, Henan Institute of Technology, Xinxiang 453002, China. E-mail: [email protected].
Abstract: As the traditional big data imputation mining process is time-consuming with low efficiency, in this paper, an incomplete big data imputation mining algorithm based on improved BP neural network was proposed. The algorithm firstly integrated into BP artificial network neural algorithm to randomly generate the initial network weight of incomplete big data, and then trained the set of weights to design an incomplete big data gene matrix. On this basis, the global search of incomplete big data information was carried out, and the big data was divided into complete and incomplete data with the search result as the core. The concept of entropy in information theory was used to perform imputation of missing values through the attribute value of the same type of complete data information. The experimental simulation proves that the incomplete big data imputation mining algorithm based on BP neural network can realize the mining of incomplete big data and improve the imputation precision of missing data.
Keywords: Incomplete big data, filling and mining algorithm, BP neural network, entropy, imputation
