RECURSIVE UPDATE SVR MODEL FOR MULTI-STEP AHEAD TIME SERIES FORECASTING STRATEGY

Abstract

This paper proposes a novel multi-step time series forecasting strategy called RU-SVR. This method iteratively updates the SVR model to enhance the accuracy and stability of forecasts amidst data fluctuations. The study conducted experiments on the VN-Index dataset from the Ho Chi Minh City Stock Exchange, comparing the performance of RU-SVR against contemporary forecasting models such as LSTM, MLP, and ConvLSTM. Experimental results indicate that RU-SVR not only achieves higher accuracy but also demonstrates greater stability in multi-step time series forecasting.