Apply Differences Grey Prediction Methods in the Selling of LOHAS

Abstract

The paper brings together the average daily sales of a Farmers' Association supermarket from May, 2010 to December, 2010. First, we examine whether the data meet the conditions of modeling, and we determine the average amount of daily sales data are in line with the GM (1,1) modeling of the capacity district. Also, it meets the conditions of the modeling prediction accuracy higher than 90%. Due to the traditional grey prediction GM (1,1) model only needs 4 data, but residual modification grey prediction GM (1,1) mode needs at least five data. Hence, the GM (1,1) model uses the raw data from May, 2010 to November, 2010 to predict values in December, 2010 to do error analysis. Also, we apply the GM (1,1) rolling test, and use the data of 5 groups, 6 groups and 7 groups to calculate six kinds of GM (1,1) prediction model, and we find the raw data with minimum error as the grey prediction GM (1,1) model. After determining the optimal grey prediction period, we apply six grey prediction methods and the optimal grey prediction period to predict the values of December, 2010 to do error analysis. We also use average error value as a criteria to select the best prediction model, and by using the top four of the best prediction model, we are able to predict next three months' growth trend.