A data clustering method and device

Abstract

The present invention provides a data clustering method and device. In the case of obtaining the uncertainty data to be clustered, based on the uncertainty probability density function of the uncertainty data, the method and device for clustering the uncertainty data are calculated. Information is required, such as uncertainty probability density function based on uncertainty data, recalculation of the preset initial centroid of the data set, uncertainty data regarded as uncertainty data relative to the expected sum of squared errors of the data set to this data set. The expected square error of the preset initial centroid recalculated by the data set and the sum of the expected square errors of the uncertainty data to the preset initial centroids of other data sets, and then determine the data set with the smallest sum of expected square errors For the target data set, divide the uncertainty data into the target data set, realize the clustering of the uncertainty data based on the uncertainty probability density function of the uncertainty data, and improve the accuracy of the uncertainty data clustering .

Description

technical field [0001] The invention belongs to the technical field of data processing, and in particular relates to a data clustering method and device. Background technique [0002] Due to inaccurate measurements, sampling errors, outdated data sources, or other reasons, data often have the nature of uncertainty (uncertainty data for short), especially in applications that require interaction with the real environment, such as mobile location services and sensor monitoring In other applications, taking the tracking of moving targets (such as vehicles or people) in mobile positioning services as an example, it is impossible to completely track the accurate instantaneous positions of all moving targets in mobile positioning services, so the position change process of each moving target is accompanied by There is uncertainty, and this uncertainty will have an impact on data management, such as data query and data clustering. [0003] There are two types of uncertainty in cur...

AI Technical Summary

Problems solved by technology

[0004] By using the EM (Expectation Maximization, maximum expectation) algorithm to solve the problem of making the mixed density suitable for uncertainty data clustering and the fuzzy C-means clustering algorithm, but these two data clustering methods do not consider the impact of uncertainty on clustering. influence, resulting in a decrease in clustering accuracy

Images