Today Xiaobian to share a case about the use of grey forecasting model method to market forecasting, interested observers together to take a look at it
Case: The Marketing Department of a brand wants to make a prediction of the future market sales scale of this category according to the known overall sales situation of the whole market of this category in the past, and hopes to get the approximate value and trend of sales in the next two years。 (The first premise is to estimate or assume that the external influence conditions are unchanged, and only consider the disturbance in the system)
The sales figures from 2012 to 2019 are as follows:
Unit: ten million yuan
In this case, the grey prediction method was adopted at that time, and the model GM (1,1) was predicted by solving the calculus equation:
The general form of the prediction model is:
The specific process is as follows:
Step 1: Using original sequence level ratio test:
Whether the detection model can adopt the gray prediction model? Although the model does not have high requirements on data, satisfying the original sequence level ratio test is also a guarantee for the accuracy of the model。It can be understood that if the original sequence ratio test is satisfied, the accuracy of the model is guaranteed and the predicted value is more accurate。
Original sequence level ratio test formula:
Through the original sequence level ratio test, it is found that:
Meet the requirements of establishment, and ensure the model accuracy is relatively accurate。
Step2Because the data meets the requirements of the gray prediction model, we can get the data results we want by modeling and predicting the code
Step3: Test and analyze the obtained model prediction results。
Unit: ten million yuan
Average relative error:
Let the variance of the residual be:
The original sequence variance is:
Of good precision。
In general, the grey prediction model only uses 3 simple steps, the model accuracy is good, and the prediction results can be better。