An Application of AdaBoost-GRU Ensemble Model to Economic Time Series Prediction

Objectives: Given the importance of accurate prediction financial time series data and their benefits in real-life, AdaBoost-GRU ensemble learning is proposed which it’s forecasting accuracy to be compared with AdaBoost-LSTM, single Long Short Term Memory (LSTM), Gated Recurrent Unit (GRU). Methods: The for Korea Composite Stock Price Index (KOSPI) obtained from Naver Finance January 2000 April 2020, Oil entire Gyeongnam region among domestic oil price Petroleum Corporation (Opinet) USD Exchange provided by Financial 2004 May 2020 were employed. analyses made using mean absolute error (MAE), squared (MSE) root (RMSE) as performance metric. Findings: Empirical results show that method outperforms all other models serve benchmarked models, three kinds used this research. This also shows have better than both AdaBoost-LSTM outperform respective GRU LSTM. Novelty/Applications: empirical study suggests ensemble-learning model a highly promising approach these data. However, another can combine AdaBoost such ConvD1 developed applied. Keywords: Price; Rate; Index; Time Series Forecasting; Algorithm;