R help - Mar 2011 - strange PREDICTIONS from a PIECEWISE LINEAR (mixed) MODEL

Hi Dears,

When I introduce an interaciton in a piecewise model I obtain some quite
unusual results.

If that would't take u such a problem I'd really appreciate an advise
from
you.

I've reproduced an example below...

Many thanks




x<-rnorm(1000)

y<-exp(-x)+rnorm(1000)

plot(x,y)
abline(v=-1,col=2,lty=2)


mod<-lm(y~x+x*(x>-1))

summary(mod)

yy<-predict(mod)

lines(x[order(x)],yy[order(x)],col=2,lwd=2)


#--lme

#grouping factor, unbalanced

g<-as.character(c(1:200))
id<-sample(g,size=1000,replace=T,
prob=sample(0:1,200,rep=T))

table(id)   #unbalanced



mod2<-lme(y~x+x*(x>-1),random=~x|id,
data=data.frame(x,y,id))

summary(mod2)


newframe<-data.frame(  #fictious id
id="fictious",
x)

newframe[1:5,]

#predictions

yy2<-predict(mod2,level=0, newdata=newframe)


lines(x[order(x)],yy2[order(x)],col="blue",lwd=2)



# add variable in the model

z<-rgamma(1000,4,6)

mod3<-lme(y~x+x*(x>-1)+z
,random=~x|id,
data=data.frame(x,y,z,id))

summary(mod3)


#new id

newframe2<-data.frame(  #fictious id
id="fictious",
x,
z)


#predict

yy3<-predict(mod3,level=0, newdata=newframe2)


lines(x[order(x)],yy3[order(x)],col="green",lwd=2)



# ADD INTERACTION  z:x



mod4<-lme(y~x+x*(x>-1)+

z+
   z:x+
         z:x*(x>-1)

   ,random=~x|id,
data=data.frame(x,y,z,id))



#predict

yy4<-predict(mod4,level=0, newdata=newframe2)


lines(x[order(x)],yy4[order(x)],col="violet",lwd=2)  #something
bizarre
                                                    #starts to happen
                                                    #in the predicted values

# they begin to jiggle around the straight line








-- 
*Little u can do against ignorance,....it will always disarm u:
is the 2nd principle of thermodinamics made manifest, ...entropy in
expansion.**....But setting order is the real quest 4 truth, ......and the
mission of a (temporally) wise dude.
*

	[[alternative HTML version deleted]]

1. Try using set.seed for better reproducibility.
2. Does this really work for you?  I get


mydf <- data.frame(x, y,id)
mod2<-lme(y ~ x + x*(x>-1), random=~x|id,
+           data=mydf)
Error in lme.formula(y ~ x + x * (x > -1), random = ~x | id, data = mydf) :
  nlminb problem, convergence error code = 1
  message = iteration limit reached without convergence (9)

Enter a frame number, or 0 to exit

1: lme(y ~ x + x * (x > -1), random = ~x | id, data = mydf)
2: lme.formula(y ~ x + x * (x > -1), random = ~x | id, data = mydf)



Kevin



On Sat, Mar 19, 2011 at 6:16 AM, Federico Bonofiglio
<bonoricus@gmail.com>wrote:
> Hi Dears,
>
> When I introduce an interaciton in a piecewise model I obtain some quite
> unusual results.
>
> If that would't take u such a problem I'd really appreciate an
advise from
> you.
>
> I've reproduced an example below...
>
> Many thanks
>
>
>
>
> x<-rnorm(1000)
>
> y<-exp(-x)+rnorm(1000)
>
> plot(x,y)
> abline(v=-1,col=2,lty=2)
>
>
> mod<-lm(y~x+x*(x>-1))
>
> summary(mod)
>
> yy<-predict(mod)
>
> lines(x[order(x)],yy[order(x)],col=2,lwd=2)
>
>
> #--lme
>
> #grouping factor, unbalanced
>
> g<-as.character(c(1:200))
> id<-sample(g,size=1000,replace=T,
> prob=sample(0:1,200,rep=T))
>
> table(id)   #unbalanced
>
>
>
> mod2<-lme(y~x+x*(x>-1),random=~x|id,
> data=data.frame(x,y,id))
>
> summary(mod2)
>
>
> newframe<-data.frame(  #fictious id
> id="fictious",
> x)
>
> newframe[1:5,]
>
> #predictions
>
> yy2<-predict(mod2,level=0, newdata=newframe)
>
>
> lines(x[order(x)],yy2[order(x)],col="blue",lwd=2)
>
>
>
> # add variable in the model
>
> z<-rgamma(1000,4,6)
>
> mod3<-lme(y~x+x*(x>-1)+z
> ,random=~x|id,
> data=data.frame(x,y,z,id))
>
> summary(mod3)
>
>
> #new id
>
> newframe2<-data.frame(  #fictious id
> id="fictious",
> x,
> z)
>
>
> #predict
>
> yy3<-predict(mod3,level=0, newdata=newframe2)
>
>
> lines(x[order(x)],yy3[order(x)],col="green",lwd=2)
>
>
>
> # ADD INTERACTION  z:x
>
>
>
> mod4<-lme(y~x+x*(x>-1)+
>
> z+
>   z:x+
>         z:x*(x>-1)
>
>   ,random=~x|id,
> data=data.frame(x,y,z,id))
>
>
>
> #predict
>
> yy4<-predict(mod4,level=0, newdata=newframe2)
>
>
> lines(x[order(x)],yy4[order(x)],col="violet",lwd=2)  #something
bizarre
>                                                    #starts to happen
>                                                    #in the predicted values
>
> # they begin to jiggle around the straight line
>
>
>
>
>
>
>
>
> --
> *Little u can do against ignorance,....it will always disarm u:
> is the 2nd principle of thermodinamics made manifest, ...entropy in
> expansion.**....But setting order is the real quest 4 truth, ......and the
> mission of a (temporally) wise dude.
> *
>
>        [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
	[[alternative HTML version deleted]]