I have the following array:
DIV RIQ EXC FX_FREQ AMPL FQMIN FQMAX FQDOM DUR
BM 1.16 14 1 6 9950 250 10200 3254.4 173.77
CP 1.23 12 1 6 9840 260 10100 3417.5 84.06
MT 2.03 14 5 9 9955 160 10115 1359.8 60.70
From it I want to run a GLM. My global template looks like this (I'm not using either RIQ or FX_FREQ):
M1<-glm(ADI ~ EXC+AMPL+FQMIN+FQMAX+FQDOM+DUR, family=poisson, data=env)
summary(M1)
When I call the summary command (M1), I get: Call:
glm(formula = ADI ~ EXC + AMPL + FQMIN + FQMAX + FQDOM + DUR,
family = poisson, data = env)
Deviance Residuals:
[1] 0 0 0
Coefficients: (4 not defined because of singularities)
Estimate Std. Error z value Pr(>|z|)
(Intercept) 5.308e+00 1.162e+02 0.046 0.964
EXC 1.406e-01 2.996e-01 0.469 0.639
AMPL -5.327e-04 1.177e-02 -0.045 0.964
FQMIN NA NA NA NA
FQMAX NA NA NA NA
FQDOM NA NA NA NA
DUR NA NA NA NA
(Dispersion parameter for poisson family taken to be 1)
Null deviance: 3.0247e-01 on 2 degrees of freedom
Residual deviance: -3.2641e-16 on 0 degrees of freedom
AIC: Inf
Number of Fisher Scoring iterations:
What bothers me is these "NA". I try to follow the analysis by creating new models, the criterion I use is: always in the new model, I do not enter the variable with the highest p value obtained in the previous model. Ex: my M2 is:
M2<-glm(ADI ~ EXC+FQMIN+FQMAX+FQDOM+DUR, family=poisson, data=env)
because no summary (M1), the AMPL variable had a higher p. I continue until I have 6 models.
In the end I want to select based on Akaike (AiCc). Then I apply the command:
AICc <- ICtab(M1,M2,M3,M4,M5,M6,type = c("AICc"),base=T, weights = TRUE,
delta = TRUE, sort = TRUE, nobs = 4)
The biggest problem occurs when I call AICc. What appears is:
AICc dAICc df weight
M1 Inf NaN 3 NA
M2 Inf NaN 3 NA
M3 Inf NaN 3 NA
M4 Inf NaN 3 NA
M5 Inf NaN 3 NA
M6 Inf NaN 2 NA
I need to sort this out. Could you help me?