Predictions And Errors Analyses Of Sea-level Series

Forecasting of time series has a great value for people in real life. Theory of time series prediction has been widely used in a lot of areas. For example, the ships’captains need to predict the wave height and the wind speed for safe navigation. The financial analysts make predictions of stock trends based on the economic data. IT specialists are monitoring network traffic passing throughout the servers in given period of time in order to predict and keep servers working in good condition.New knowledge is the renew product of previous known. As the matter of fact, time-series prediction is also based on the observation historical data. The theory of time series analysis was established many years ago. Recently, researchers have proposed many better and more accurate methods. However, even the best of these models cannot always work with some kinds of data. There are also some errors occurring in data prediction. Sometimes the error term value seems small, but it has a great impact on the results, such as in the aviation industry and electronics industry. Even minor errors will have serious results for practical applications in these areas. Therefore, this article focused on the error term analysis of sea level data in time series prediction.The article contains two parts:one is about methods of time series prediction of sea level data, the other is about the quantitative error term analysis of time series prediction of sea level data. Some experts get the conclusion that the longer the sample size is, the smaller the prediction error becomes. However, if the error term is given, then how long the sample size is? Few scientists worked on this topic. This paper presented measurement technique of the correlation between error term and sample size of sea level data. We used sea level data for forecasting and calculated errors under different sample lengths. Finally, we presented the function of error item and sample size. This function gives us a numerical calculation method of correlation between error term and sample size of sea level data. In addition, we compared the prediction errors under the conditions of different values of the Hurst parameter.