A data processing method, a data processing apparatus, and a computing device cluster
By performing anomaly detection and feature analysis on aggregated sequences in cloud service scenarios, and correcting data in abnormal intervals, the problem of directly repairing the impact of abnormal data on trends in aggregated sequences is solved, thereby improving prediction accuracy.
CN122413104APending Publication Date: 2026-07-17HUAWEI TECH CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUAWEI TECH CO LTD
- Filing Date
- 2025-01-16
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
In cloud service scenarios, abnormal data caused by accidental factors are not directly related to the overall trend in time series. Directly repairing abnormal data in aggregated series may incorrectly change the trend of subsequences and affect the accuracy of prediction.
Method used
By performing anomaly detection on the aggregated sequences, abnormal intervals are identified, and the characteristic information of each time series is analyzed to correct the data in the abnormal intervals, ensuring that the trend of the time series changes before aggregation is not affected.
Benefits of technology
It improves the accuracy of time series forecasting and avoids the impact of directly repairing anomalous data in aggregated sequences on the original time series trend.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN122413104A_ABST
Abstract
提供了一种数据处理方法、数据处理装置、以及计算设备集群。该方法包括：对聚合序列进行异常检测，确定第一异常区间，该聚合序列根据多个时间序列得到；对于多个时间序列中的第一时间序列，根据第一时间序列中位于第一异常区间中的目标数据对应的特征信息和第一时间序列中除目标数据之外的其他数据对应的特征信息，确定目标数据是否为异常数据；在确定目标数据为异常数据的情况下，修正目标数据。该方法在确定聚合序列存在异常后，对聚合前的各个时间序列进行分析，并修正时间序列中的异常数据，可以消除该时间序列的偶然因素对聚合序列的影响。
Need to check novelty before this filing date? Find Prior Art