A Real-Time Computer Network Trend Analysis Algorithm Based on Dynamic Data Stream in the Context of Big Data

The extraction of effective trends can provide early warning, status assessment and decision support for monitoring objects. Traditional curve trend analysis algorithms include sliding window (SW) algorithm and extrapolation online data segmentation (OSO) algorithm. Compared with conventional least squares method, the overall least-squares method has a higher accuracy of straight line fitting. In addition, there is no limit to the maximum length of the sliding window for the SW algorithm. When the threshold of the detection point is relatively large, the length of the window may be long. The OSD algorithm defines the minimum sliding window length, so that the mutation point within the minimum sliding window cannot be detected. Aiming at the shortcomings of the SW algorithm and the OSD algorithm, a new data stream trend analysis method is proposed. This method adopts the overall least squares method to fit the data stream segmentally to improve the precision of the trend analysis. In addition, it also proposes a variable sliding window. The algorithm solves the fixed window problem of the SW algorithm and the OSD algorithm to achieve a reasonable segmentation of the data stream.