Dear Prof. Thomas Yee
I$B!G(Bm very interested in your R program VGAM.
I tried below your data:
# Nonparametric proportional odds model
data(pneumo)pneumo = transform(pneumo,
let=log(exposure.time))vgam(cbind(normal,mild,severe) ~ s(let),
cumulative(par=TRUE), pneumo)
However, the results by Version of VGAM are different;
----------The result by Version 0.7-7 -----------------------------
Call:
vgam(formula = cbind(V1, V2, V3) ~ s(V4), family = cumulative(par = TRUE),
data = train)
Number of linear predictors: 2
Names of linear predictors: logit(P[Y<=1]), logit(P[Y<=2])
Dispersion Parameter for cumulative family: 1
Residual Deviance: 2.9184 on NaN degrees of freedom
Log-likelihood: -203.2200 on NaN degrees of freedom
Number of Iterations: 6
DF for Terms and Approximate Chi-squares for Nonparametric Effects
Df Npar Df Npar Chisq P(Chi)
(Intercept):1 1
(Intercept):2 1
s(V4) 1
----------The result by Version 0.7-3 -----------------------------
Call:
vgam(formula = cbind(V1, V2, V3) ~ s(V4), family = cumulative(par = TRUE),
data = train)
Number of linear predictors: 2
Names of linear predictors: logit(P[Y<=1]), logit(P[Y<=2])
Dispersion Parameter for cumulative family: 1
Residual Deviance: 2.37107 on 10.368 degrees of freedom
Log-likelihood: -202.9463 on 10.368 degrees of freedom
Number of Iterations: 7
DF for Terms and Approximate Chi-squares for Nonparametric Effects
Df Npar Df Npar Chisq P(Chi)
(Intercept):1 1
(Intercept):2 1
s(V4) 1 2.6 1.95553 0.50986
I think that the result by Version 0.7-3 is right.
Please teach me if my result is right.
Yours sincerely,
Prof. M.Tsujitani
Osaka Electro-Communication University
[[alternative HTML version deleted]]
