用S-PLUS试了下,结果如下:
y=FARIMA(x,p.range=c(0,2),q.range=c(0,2),mmax=0)
结果:
p = 0 q = 0
p = 0 q = 1
p = 0 q = 2
p = 1 q = 0
p = 1 q = 1
p = 1 q = 2
p = 2 q = 0
p = 2 q = 1
p = 2 q = 2
summary(y)
结果:
Call:
FARIMA(x = x, p.range = c(0, 2), q.range = c(0, 2), mmax = 0)
Coefficients:
Value Std. Error t value Pr(>|t|)
d -0.0848 0.0234 -3.6194 0.0003
AR(1) 0.7257 0.0189 38.3355 0.0000
MA(1) 0.4391 0.0087 50.4501 0.0000
Information Criteria:
log-likelihood BIC
-24887.50 49804.55
Residual scale estimate: 0.8981
total residual
Degree of freedom: 18978 18974
BIC of all models estimated:
q=0 q=1 q=2
p=0 49956.33 49955.87 49917.32
p=1 49953.95 49804.55 49813.03
p=2 49887.75 49814.71 49807.89
模型定阶为p=1、q=1,是根据BIC准则确定的。但感觉其对参数d的估计不好,d=-0.0848,d应该在0~0.5的范围内才合适。用R计算相同阶数时对应的d值为0.2481836。与S-PLUS估计的参数存在明显差别。