Time Series Data Acquisition And Application

In this information age,with the development of technologies such as Internet of Things,Cloud Computing,Social Networking,and Artificial Intelligence,massive information is constantly being produced,there are many time series data in those massive dates.There are many valuable information behind the ever-increasing time series data,however,without relevant theory and technology,we can do nothing about it,how to find valuable and useful information from those data become an urgent need for people today.So,more and more researchers invest themselves into the research of prediction time series.This article focuses on improving the prediction accuracy of time series.First,this article analyzes the importance of quality data for time series,then get quality data by Researching and Improving the data acquisition method of Abnormal magnetic flux leakage detection data acquisition in power system and Campus Network Social Website User Access Data Collection from the perspective of data acquisition,and provides effective data protection for further research on time series forecasting.For time series forecasting,this article has improved from the following two aspects:(1)Considering that traditional models only consider information under a single time granularity when predict time series,and has low prediction accuracy,this article proposes a model of Gaussian Process Regression(GPR)based on the feature of multi-time granularity.(2)Because of the uncertainty in time series changes,prediction accuracy is difficult to achieve the desired effect,we propose a combination forecast model of GPR based on the feature of multi-time granularity which use Discrete Wavelet Transformation(DWT)and Weighted Nearest Neighbors(WNN).