Taylor series approximation and unbiased GM(1,1) based hybrid statistical approach for forecasting daily gold price demands

Abstract

Purpose – The time series analysis is an essential methodology which comprises the tools for analyzing the time series data to identify the meaningful characteristics for making future ad-judgments. The purpose of this paper is to propose a Taylor series approximation and unbiased GM(1,1) based new hybrid statistical approach (HTS_UGM(1,1)) for forecasting time series data under the poor, incomplete and uncertain information systems in a short period of time manner. Design/methodology/approach – The gray forecasting is a dynamical methodology which can be classified into different categories based on their respective functions. The new proposed methodology is made up of three different methodologies including the first-order unbiased GM(1,1), Markov chain andTaylorapproximation. Inadditionto that,twodifferenttraditionalgray operational mechanismsinclude GM(1,1) and unbiased GM(1,1) used as the comparisons. The main objective of this study is to forecast gold price demands in a short-term manner based on the data which were taken from the Central Bank of Sri Lanka from October 2017 to December 2017. Findings – The error analysis results suggested that the new proposed HTS_UGM(1,1) is highly accurate (less than 10 percent) with lowest RMSE error values in a one head as well as weakly forecasting’s than separate gray forecasting methodologies. Originality/value – The findings suggested that the new proposed hybrid approach is more suitable and effective way for forecasting time series indices than separate time series forecasting methodologies in a short-term manner.