A Novel Modeling Method of Extended Grey EGM(1,1,∑e^(ck))
Model and Its Application in Predictions
Maolin Cheng, Bin Liu
2023, 35(4):
55-75.
Asbtract
(
167 )
Related Articles |
Metrics
In the grey models, the GM(1,1) model is an important type of prediction model.
The traditional grey GM(1,1) model has good prediction results in the case the
original data show exponential variations at a slow rate. However, in practical
problems, although showing exponential variations or approximately exponential
variations, original data vary very fast sometimes. In these cases, the traditional grey
GM(1,1) model tends to have poor prediction accuracy, mainly because the data fails
to meet the laws presented by the traditional model. Therefore, the paper makes
improvements in the following two aspects: first, the paper transforms the traditional
accumulated generating sequence of original data; second, the paper extends the
traditional grey model's structure, i.e., building a grey EGM(1,1,∑e^(ck)) model. The
paper offers the parameter optimization method of the grey EGM(1,1,∑e^(ck)) model.
Using the novel modeling method proposed, the paper builds the grey
EGM(1,1,∑e^(ck)) models for China's total electricity consumption and China's
GDP per capita, respectively, in the final section. Results show that the models built
with the proposed modeling method have high simulation precision and prediction
precision.
