Classification and Forecasting for Enterprise Data

摘要

There is an increasing interest in classification and forecasting for enterprise data. Besides many similarity or dissimilarity measures, Support Vector Machines have also been used for time series classification. We classify the industry data with autocorrelation function-distance, an automatic adaptive dissimilarity index and Support Vector Machines respectively. Then we make forecasting for different class. Every class is fitted with particular season model according to Akaike Information Criteria. The orders of the particular model are estimated. A comparative study is presented. It is proposed that an automatic adaptive dissimilarity index outperforms autocorrelation function distance and Support Vector Machines.