What Is Exponential Smoothing?

The exponential smoothing method is actually a special weighted moving average method. Its characteristics are: First, the exponential smoothing method further strengthens the effect of recent observations on the predicted value during the observation period. The weights given to the observations at different times are different, thereby increasing the weight of the recent observations, Forecasts can quickly reflect actual market changes. The weights are reduced in equal series. The first term of this series is the smoothing constant a, and the common ratio is (1-a). Second, the exponential smoothing method has scalability for the weights assigned to the observations, and can take different values of a to change the rate of change of the weights. If a is a small value, the weight changes more quickly, and the recent change trend of the observed value can be more quickly reflected in the exponential moving average. Therefore, using the exponential smoothing method, you can choose different values of a to adjust the uniformity of the time series observations (that is, the smoothness of the trend change). ^[1]

Exponential smoothing is a method commonly used in production forecasting. It is also used to forecast the short-term economic development trend. Among all forecasting methods, exponential smoothing is the most used one. Simple full term

According to the different smoothing times, the exponential smoothing method is divided into: one exponential smoothing method,

The upward or downward trend in the data collected over a period of time will cause the index forecast to lag behind actual demand. Add trends through trend adjustment

Take a software company A as an example, give the historical sales data from 2000 to 2005, substitute the data into an index smoothing model, and predict the sales in 2006 as the basis for the sales budget preparation.

According to the empirical judgment method, the sales time series of Company A from 2000 to 2005 fluctuated greatly, and the long-term trend changed greatly, showing a clear and rapid upward trend. It is appropriate to choose a large value, which can be selected between 0.5 and 0.8 In order to make the prediction model more sensitive, take 0.5, 0.6, and 0.8 to test respectively in combination with the trial algorithm. After the first exponential smoothing, the series scatter plot shows a straight line trend, so the second exponential smoothing method can be used.

According to the mean of the squared deviation (MSE) is the smallest, that is, the sum of the square of the difference between the actual value and the predicted value of each period divided by the total number of periods, the minimum value is used to determine the standard value of . 1445.4; when = 0.8, MSE2 = 10783.7; when = 0.5, MSE3 = 1906.1. Therefore, we choose = 0.6 to forecast the sales in the fourth quarter of 2006.