Applying Support Vector Machine to Time Series Prediction in Oracle

Abstract

Using oracle data mining option(ODM) and the time series stored in oracle database, the SVM (support vector machines) model can be used to predict the future value of the time series. To build SVM model, firstly the trend in time series must be removed, and the target attribute should be normalized. secondly the size of the time window in which include all the lagged values should be determined, thirdly the machine learning method is used to construct SVM prediction model according to the time series data. Comparing with the traditional time series prediction model, SVM prediction models can reveal non-linear, non-stationary and randomness of the time series, and have higher prediction accuracy.