Research On Incomplete Data Filling Algorithm For Internet Of Things

With the large-scale application of the technology of the Internet of Things, the data incompleteness of the Internet of Things is increasing prominent. So-called data completeness is the acquisition of data attribute value missing, the terminal is operating abnormalities caused by. Incomplete data of the Internet of Things has brought great difficulties to data fusion, data mining, seriously block the application of the Internet of Things data. Therefore the incomplete data filling is an important topic.For incomplete data of the Internet of Things processing, mainly did two researches, one is data filling algorithm and the other is rough set theory. Major research in the rough set theory attribute attribute reduction and granulocyte-related knowledge with traditional knowledge to deal with incomplete information system to the Internet of things incomplete data abstraction, and propose a new definition of the incomplete information system by improved attribute reduction algorithm based on the power graph. Many classic methods such as incomplete data filling algorithm mean, probability, linear regression forecast, based on the decision tree filling method, but these methods of treatment using incomplete data is a unified approach, is not suitable for processing large scale data of Things, this paper presented incomplete data filling algorithm to distinguish between the important attributes and non-important attribute. Packet probability-based approach filled for non-important attributes, and similarity-based method filled important attributes. This method not only can guarantee the accuracy of the important attribute missing values, but also can reduce the computational complexity of the non-important attributes.In this paper, the data collected by the experimental platform for the Internet of Things as a data source simulation experiments, experiments show that the proposed method to the Internet of Things effectively filled the incomplete data, superior accuracy than probability or means.