An unbiased GM(1,1)-based new hybrid approach for time series forecasting

Abstract

Purpose The time series forecasting is an essential methodology which can be used for analysing time series data in order to extract meaningful statistics based on the information obtained from past and present. These modelling approaches are particularly complicated when the available resources are limited as well as anomalous. The purpose of this paper is to propose a new hybrid forecasting approach based on unbiased GM(1,1) and artificial neural network (UBGM_BPNN) to forecast time series patterns to predict future behaviours. The empirical investigation was conducted by using daily share prices in Colombo Stock Exchange, Sri Lanka. Design/methodology/approach The methodology of this study is running under three main phases as follows. In the first phase, traditional grey operational mechanisms, namely, GM(1,1), unbiased GM(1,1) and nonlinear grey Bernoulli model, are used. In the second phase, the new proposed hybrid approach, namely, UBGM_BPNN was implemented successfully for forecasting short-term predictions under high volatility. In the last stage, to pick out the most suitable model for forecasting with a limited number of observations, three model-accuracy standards were employed. They are mean absolute deviation, mean absolute percentage error and root-mean-square error. Findings The empirical results disclosed that the UNBG_BPNN model gives the minimum error accuracies in both training and testing stages. Furthermore, results indicated that UNBG_BPNN affords the best simulation result than other selected models. Practical implications The authors strongly believe that this study will provide significant contributions to domestic and international policy makers as well as government to open up a new direction to develop investments in the future. Originality/value The new proposed UBGM_BPNN hybrid forecasting methodology is better to handle incomplete, noisy, and uncertain data in both model building and ex post testing stages.

A review of artificial intelligence quality in forecasting asset prices

“…In the daily timeframe, Rasel et al (2015) achieved the best result with a MAPE of 0.01% for S&P500 using an SVM‐based model. Rathnayaka et al (2016) also achieved a high‐quality MAPE for the ASPI index, below 0.04%. Although these results are considered excellent, other studies from the same period produced lower results, such as Baek and Kim (2018) and Nadhem et al (2015), who obtained average percentage errors above 1% for the same index examined by Rasel et al (2015).…”

Predicting of aging population density by a hybrid grey exponential smoothing model (HGESM): a case study from Sri Lanka

“…The concept of solving complex problems (CPS) based on achievements and workplace performances has been recognized as one of the major skills. The miscellaneous types of methodologies have been developed in the literature to analyze complex real-world problems; especially for arranging and maintaining a set of higher-order skills that are required in strategic planning, building and testing coming up with new hypotheses (Rathnayaka et al. , 2016b).…”

“…According to the literature, different methodologies such as the Malthus model (Wang, 2006), the Logistic model (Liu et al ., 2022a, b), the linear regression model (Rathnayaka and Seneviratna, 2020; Rathnayaka et al. , 2016a, b), the neural network model (Wang, 2006) and the Grey system model (Li et al. , 2019) have developed to predict population rates based on their own assumptions, characteristics and conditions.…”

A Statistical Analysis of the Colombo Stock Returns

“…Chandrasekara, Mammadov, and Tilakaratne (2016) analysed the Colombo stock indices only to illustrate the fit of a multivariate scaled t distribution. Rathnayaka, Seneviratna, Wei, andArumawadu (2016a, 2016b) used k-means clustering, grey mechanism, autoregressive integrated moving average processes and artificial neural network in their analysis. But these methods suppose that data are normally distributed.…”