search for


Financial time series forecasting using AdaBoost-GRU ensemble model
Journal of the Korean Data & Information Science Society 2021;32:267-81
Published online March 31, 2021;
© 2021 Korean Data and Information Science Society.

Nae Won Kwak1 · Dong Hoon Lim2

12Department of Information Statistics, Gyeongsang National University
Correspondence to: 1Undergraduate, Department of Information Statistics, Gyeongsang National University, Jinju 52828,Korea.
2Corresponding author: Professor, Department of Information Statistics/Bio & Medical Big Data, Gyeongsang National University, Jinju 52828, Korea. E-mail:
Received November 25, 2020; Revised December 29, 2020; Accepted January 3, 2021.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
In general, forecasting an financial time series is very difficult due to non-linearity and irregularity. In this paper, we propose a hybrid ensemble learning approach that combines the AdaBoost algorithm and GRU model for financial time series forecasting. Here, the GRU model is a modified structure of a recurrent neural network (RNN) widely used for time series forecasting along with a long short term memory (LSTM) model. We evaluated the proposed model with two financial time series data: KOSPI data, and won/dollar exchange rate data. As a result of performance tests, the proposed AdaBoost-GRU ensemble showed better performance than ARIMA model, single LSTM model, single GRU model, and Adaboost-LSTM ensemble in three scales: MAE, MSE and RMSE. In addition, the proposed AdaBoost-GRU ensemble was found to be fast in terms of processing speed with the Adaboost-LSTM model.
Keywords : AdaBoost-GRU ensemble, deep learning, financial time series.