Abstract: This paper presents an impuation algorithm based on learning for incomplete big data.The proposed algorithm establishs a novel auto-encoder,called imputation auto-encoder,and then builds a deep imputation network model to analyze the deep features of incomplete big data and to calculate network parameters based on drill training ideas and back-propagation algorithm.Finally,the deep imputation network is used to impute the missing values.Experimental results show that the proposed algorithm can effectively improve the imputation accuracy for incomplete big data.