Non linear autoregressive model
LSTAR model
Coefficients:
Low regime:
const.L phiL.1 phiL.2 phiL.3 phiL.4
4.994274112 0.618301630 -0.005824329 0.250343820 0.143844290
High regime:
const.H phiH.1 phiH.2 phiH.3 phiH.4
-5.51010419 -0.05233469 0.02415548 -0.11617584 -0.09408755
Smoothing parameter: gamma = 84.77
Threshold
Variable: Z(t) = + (1) X(t) + (0) X(t-1)+ (0) X(t-2)+ (0) X(t-3)
Value: -23.82
Residuals:
Min 1Q Median 3Q Max
-61.231126 -6.613737 0.048184 6.008438 103.466675
Fit:
residuals variance = 189.1, AIC = 7185, MAPE = 196.1%
Coefficient(s):
Estimate Std. Error t value Pr(>|z|)
const.L 4.9943e+00 2.6855e+00 1.8597 0.062924 .
phiL.1 6.1830e-01 7.5018e-02 8.2421 2.22e-16 ***
phiL.2 -5.8243e-03 6.9857e-02 -0.0834 0.933553
phiL.3 2.5034e-01 7.9909e-02 3.1329 0.001731 **
phiL.4 1.4384e-01 6.2917e-02 2.2862 0.022240 *
const.H -5.5101e+00 2.7146e+00 -2.0298 0.042378 *
phiH.1 -5.2335e-02 8.2340e-02 -0.6356 0.525043
phiH.2 2.4155e-02 7.7972e-02 0.3098 0.756715
phiH.3 -1.1618e-01 8.6706e-02 -1.3399 0.180283
phiH.4 -9.4088e-02 6.9650e-02 -1.3509 0.176739
gamma 8.4769e+01 1.1880e+03 0.0714 0.943115
th -2.3816e+01 2.6542e-01 -89.7307 < 2.2e-16 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Non-linearity test of full-order LSTAR model against full-order AR model
F = 2.4322 ; p-value = 0.045775
Threshold
Variable: Z(t) = + (1) X(t) + (0) X(t-1)+ (0) X(t-2)+ (0) X(t-3)