A hybrid model to improve the capabilities of forecasting based on GRA and ANN theories

In this study, the Grey Relational Analysis (GRA) method is combined with artificial neural networks (ANN) model to create an automatic stock forecasting mechanism. In the proposed approach, the attributes of quarterly datum with the same category are gathered into a specific financial ratio by the GRA method. The categorical data is then input to an ANN model to forecast the future trends of the collected data over the next quarter or half-year period. The validity of the proposed approach is demonstrated using electronic stock data extracted from the financial database maintained by the Taiwan Economic Journal (TEJ). The hybrid forecasting model using five GRA methods and ANN model is employed to identify the feasibilities which achieve the criteria of predictive ability. Then, the predictive ability obtained using the proposed hybrid model are compared with those of a ANN prediction method reduced the attributes of forecast data by GM(1, N). It is found that the proposed method not only need a less number of neurons than the ANN combined with GM(1, N) method, but also has a greater forecasting accuracy.