Research And Application Of Sky Drive Capacity Trend Prediction Algorithm Based On Time Sequeue

With the development of the Internet and the progress of information technology, people are getting higher and higher requirements for storage, whether it is storage capacity or storage convenience requirements. Is a concomitant effect of storage has been from the physical storage such as U disk, mobile hard disk, such as not very convenient way gradually to the virtual memory, such as the development of network disk. Increasing the number of Internet users, Sky Drive capacity has become more and more. For Sky Drive providers to ensure the stability Sky Drive capacity has become a crucial task. At the same time, we need to start from the interests of the company, can not appear idle waste of resources. Therefore, how to accurately grasp the trend of the network disk capacity, according to changes in the capacity to make decisions is undoubtedly the urgent need for various network disk provider.In the network disk capacity data analysis phase. Firstly, systematically on the survey data were research and analysis, in view of the characteristics of the data, using data mining theory knowledge, the common trend prediction algorithm for learning and research, and comparison and testing of each algorithm, finally, according to the actual situation of the using median filter on the specific implementation of data of salt and pepper noise denoising, using moving average method for Gaussian noise denoising, using the two methods to carry on the data pretreatment. Secondly, using t test with normal distribution 3sigma theory of outlier detection, to extract predictions about the effects of outliers, and skip over these points. At last, the linear fitting method for the linear of the processed data were fitted, and put forward and its practical application to meet the trend of time series based on prediction algorithm.In the capacity of prediction analysis stage, the extensive experiments, and the experimental results were compared, in the selection of some parameters has also been a large number of experiments, so as to ensure the reliability of the overall algorithm. Finally, using the volume trend prediction algorithm, according to the actual data were a period of experiments and statistical, compares the statistics prediction result and the real result and error in line with expectations.