Two-phase hybridisation using deep learning and evolutionary algorithms for stock market forecasting. (24th December 2021)

• Record Type: Journal Article
• Authors: Kumar, Raghavendra; Kumar, Pardeep; Kumar, Yugal
• Is Part Of: International journal of grid and utility computing. Volume 12:Number 5/6(2021)
• Page Start: 573
• Page End: 589
• Publication Date: 2021-12-24
• Subjects: hybrid model -- ARIMA -- auto regressive integrated moving average -- LSTM -- long short-term memory -- ABC -- artificial bee colony; Electronic data processing -- Distributed processing -- Periodicals; Electronic commerce -- Management -- Computer programs -- Periodicals; 004.605
• Languages: English
• ISSNs: 1741-847X

Abstract

In this paper, a two-phase hybrid model is proposed for stock market forecasting using deep learning approach and evolutionary algorithms. In the first phase of hybridisation, Auto Regressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) are combined to compose linear and non-linear features of the data set. In the second phase, an improved Artificial Bee Colony (ABC) algorithm using Differential Evolution (DE) is used for the hyperparameter selection of proposed hybrid LSTM-ARIMA model. In this paper, experiments are performed over 10 years of the data sets of Oil Drilling & Exploration and Refineries sector of National Stock Exchange (NSE) and Bombay Stock Exchange (BSE) from 1 September 2010 to 31 August 2020. Obtained result demonstrates that the proposed LSTM-ARIMA hybrid model with improved ABC algorithm has superior performance than its counterparts ARIMA, LSTM and hybrid ARIMA-LSTM benchmark models.