Question

我在R中使用HoltWinters()来适应模型。我无法理解为什么该模型为alpha，beta和gamma提供零值。在这种情况下该怎么办？以下是数据和代码。提前致谢。

x<-c(2835,  2970,   2565,   2970,   2700,   4995,   3375,   
2430,   2295,   3375,   3780,   1755,   2970,   2835,   
1890,   1485,   3780,   2430,   3780,   1215,   3915,   
1485,   3375,   3510,   2970,   2700,   2025,   3105,
3645,   2721,   1215,   4995,   2430,   2700,   1620,   
2025,   2430,   3780,   2295,   4860,   1350,   3645,   
6885,   2295,   1350,   2565,   2835,   2970,   2970,   
2970,   4185,   3105,   1080,   2565,   2295,   3105,
1890,   3386,   3375,   3105,   2565,   2970,   5265,   
5400,   2565,   1890,   3240,   4050,   2160,   4725,   
5805,   6615,   2565,   4455,   1620,   2025,   1620,   
2835,   1485,   4455,   3510,   1215,   2025,   3510,
3510,   2160,   1755,   1620,   4050,   2295,   2970,   
2108,   3510,   4725,   2565,   5535,   2025,   1080,   
675,    135,    2700,   3904,   2835,   3105,   1755,   
2565,   1755,   2970,   1080,   1755)

x_ts<-ts(x,frequency=52)
hw<-hw<-HoltWinters(x_ts,seasonal="multiplicative")

Output:
hw

Holt-Winters exponential smoothing with trend and multiplicative seasonal component.

Call:
HoltWinters(x = x_ts, seasonal = "multiplicative")

Smoothing parameters:
alpha: 0
beta : 0
gamma: 0

Coefficients:
        [,1]
a   3108.2165107
b      3.6027043
s1     1.0734921
s2     0.9859404
s3     0.8155082
s4     0.9450738
...

Answer 1

您的数据不足以适应大频率，因此您可以降低频率意味着您将获得非零alpha值。例如：

x_ts<-ts(x,frequency=5)

在拟合季节模型时，HoltWinters（）中的参数值为零

1 个答案: