Research Information System
2016 3rd International Conference on Mechanical, Electronics and Computer Engineering (CMECE 2016), New York, United States Of America, 7 - 09 January 2016, pp.38-39, (Summary Text)
In literature there are many methods for forecasting time
series and grey system theory based methods are one of them. Grey
system based models need little origin data, have simple calculate
process and higher forecasting accuracy with lower estimation errors,
they have been widely used in the prediction of a lot of research fields.
Grey theory, originally developed by Deng (1982). Grey System Theory
focuses on model uncertainty and information insufficiency in analyzing
and understanding systems. Recent years, researchers have worked to
propose new models that incorporate the grey prediction theory with
theories to enhance the forecasting precision. Common part of these
purposed models to focus on improving the grey predictive abilities to
obtain higher forecasting accuracy. Genetic algorithm is one of the most
known metaheuristic techniques. It is population based random search
procedure that starts with random initial population. In this study
genetic algorithm metaheuristic used to supplement the grey prediction
model and determine the grey model coefficients. Also the grey
forecasting models have been developed and compared with genetic
algorithm implantation.The analysis has been done for the period of
2011-15 of the advertising expenditure in the print media namely
magazines and newspapers in the Sultanate of Oman, and predictions
for the year 2016. Genetic algorithm based grey prediction models used
and compared with grey prediction models. Genetic algorithm
outperformed basic grey prediction models. Purposed methods used for
generating forecast models for Oman Media Outlook dataset. Time
series include net advertising spends with type Newspapers, Magazines
and Total spends. We generated 4 models for these time series and used
GM(1,1) and BGM(1,1) methods and their GA integrated models namely
GA-GM(1,1) and GA-BGM(1,1). Models’ results of three type ad spend
dataset show that GA integrated GA-GM(1,1) and GA-BGM(1,1) models
make the forecast with lower errors than GM(1,1) and BGM(1,1)
models. From analysis results it can be seen that there is a steady increase of expenditure in both the media. GA and BGM models show
that the increase is to a total of 80,72340248 - 81,05404513 and
81,52535954 – 82,39371685 respectively. The upward trend represents
increasing creativity from the producers and the changing tastes and
requirements of the readers.