Search results

The purpose of this paper is to propose a new grey system model used for prediction.

It had been proven that the GM(1,1) model is a biased exponential model, the model is fit for non‐negative raw data, which accord with or basically accord with the exponential form and do not have a quick growth rate. Based on the results, an unbiased GM(1,1) model was proposed. With the method of transforming every datum of raw data sequence into its 2‐th root, a new data sequence from the raw data sequence can be produced. The new data sequence is used to establish an unbiased GM(1,1) model and statistical experiments and a practical example in load forecasting are given in the paper.

The results of statistical experiments and a practical example in load forecasting show the proposed method is effective in increasing the accuracy of the model.

The model exposed in the paper can be used for constructing models of prediction in many fields such as agriculture, electric power, IT, transportation, economics, management, etc.

The paper succeeds in proposing a modified unbiased GM(1,1) model that has high accuracy. The model is applied to the field of load forecasting and the results show the model is better than the unbiased GM(1,1) model. The model proposed has great theoretical and practical value.

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.

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.

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.

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.

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.