Re: [R] Problem of intercept?
Dear Peter, Bert and Timothy,
Really appreciated your help. I guess Peter's comments were the nearest of what
I was trying to do.
I have tried but it still does not work. I mean the graph is showing the
coefficients I am not expecting !
I will try to explain.
Starting from the equation : Y = b0 + b1*x1 where bo and b1 are the regression
beta coefficients representing the intercept and the slope, respectively.
Suppose that, I wish to investigate the differences in Y between males and
females. Based on the gender variable, I can create a new dichotomous/dummy
variable that takes the value :
1 if a person is male
0 if a person is female
and use this variable as a predictor in the regression equation, leading to the
following model :
b0 + b1 if person is male
b0 if person is female
the coefficients can be interpreted as follow:
1. b0 is the average Y among female
2. b0 + b1 is the average Y among male
3. and b1 is the average difference in Y between males and females
The graph with my R code is showing the coefficients I was expecting for
multiple regression but not for simple regression. I mean I am expecting to see
b0 + b1 for coefficients in the plot for simple regression and the plot is
showing me b1 for simple regression ! Is there a solution to my problem?
Best,
Le mercredi 22 février 2023 à 10:59:38 UTC+1, peter dalgaard
a écrit :
Not sure what you are trying to do here.
The immediate issue is that you are getting 'y' on the RHS, because that is the
1st column in Dataset. So "for (i in 2:3)" might be closer to intention.
However, a 0/1 regresson with no intercept implies that the mean for the "0"
group is zero, and with two regressors that the mean is zero for the (0,0)
group. Looking at the data, this is quite clearly not the case.
I suppose you may have intended to fit the models _with_ the intercept and then
_ignore_ the intercept for plotting purposes, i.e. lm(y~x11+x12,
Dataset)$coef[-1], etc.?
(Also, I suspect that you don't actually have y=7 and y=867 in the dataset.)
-pd
> On 21 Feb 2023, at 22:33 , varin sacha via R-help
> wrote:
>
> Dear R-experts,
>
> Here below my R code working with quite a few warnings.
> x11 and x12 are dichotomous variable (0=no and 1=yes). I substract 1 to
> ignore intercept.
> I would like not to ignore intercept. How to modify my R code because if I
> just remove -1 it does not work?
>
>
> y= c(32,45,65,34,23,43,65,76,87,98,7,867,56,45,65,76,88,34,55,66)
> x11=c(0,1,1,0,0,1,1,1,0,0,1,0,0,1,0,0,1,1,0,1)
> x12=c(0,1,0,1,0,1,1,0,1,1,0,0,1,1,1,0,0,1,0,0)
>
> Dataset=data.frame(y,x11,x12)
>
> a=lm(y~x11+x12-1,Dataset)$coef
> b=NULL
> for(i in c(1:2)) {
> f=formula(paste('y~',names(Dataset)[i],-1))
> b=c(b,lm(f,Dataset)$coef)
> }
> coef=data.frame(rbind(a,b))
> coef$Model=c('Multi','Single')
> library(reshape2)
> coef.long<-melt(coef,id.vars="Model")
>
> library(ggplot2)
> ggplot(coef.long,aes(x=variable,y=value,fill=Model))+
> geom_bar(stat="identity",position="dodge")+
> scale_fill_discrete(name="Model",
> labels=c("Multiple", "Simple"))+
> labs(title =paste('La différences des coefficients
> entre la régression multiple et simple'),
> x="Models",y="Coefficient")+
> coord_flip()
>
>
>
>
> __
> R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
> 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.
--
Peter Dalgaard, Professor,
Center for Statistics, Copenhagen Business School
Solbjerg Plads 3, 2000 Frederiksberg, Denmark
Phone: (+45)38153501
Office: A 4.23
Email: pd@cbs.dk Priv: pda...@gmail.com
__
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
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.
Re: [R] Problem of intercept?
Not sure what you are trying to do here.
The immediate issue is that you are getting 'y' on the RHS, because that is the
1st column in Dataset. So "for (i in 2:3)" might be closer to intention.
However, a 0/1 regresson with no intercept implies that the mean for the "0"
group is zero, and with two regressors that the mean is zero for the (0,0)
group. Looking at the data, this is quite clearly not the case.
I suppose you may have intended to fit the models _with_ the intercept and then
_ignore_ the intercept for plotting purposes, i.e. lm(y~x11+x12,
Dataset)$coef[-1], etc.?
(Also, I suspect that you don't actually have y=7 and y=867 in the dataset.)
-pd
> On 21 Feb 2023, at 22:33 , varin sacha via R-help
> wrote:
>
> Dear R-experts,
>
> Here below my R code working with quite a few warnings.
> x11 and x12 are dichotomous variable (0=no and 1=yes). I substract 1 to
> ignore intercept.
> I would like not to ignore intercept. How to modify my R code because if I
> just remove -1 it does not work?
>
>
> y= c(32,45,65,34,23,43,65,76,87,98,7,867,56,45,65,76,88,34,55,66)
> x11=c(0,1,1,0,0,1,1,1,0,0,1,0,0,1,0,0,1,1,0,1)
> x12=c(0,1,0,1,0,1,1,0,1,1,0,0,1,1,1,0,0,1,0,0)
>
> Dataset=data.frame(y,x11,x12)
>
> a=lm(y~x11+x12-1,Dataset)$coef
> b=NULL
> for(i in c(1:2)) {
> f=formula(paste('y~',names(Dataset)[i],-1))
> b=c(b,lm(f,Dataset)$coef)
> }
> coef=data.frame(rbind(a,b))
> coef$Model=c('Multi','Single')
> library(reshape2)
> coef.long<-melt(coef,id.vars="Model")
>
> library(ggplot2)
> ggplot(coef.long,aes(x=variable,y=value,fill=Model))+
> geom_bar(stat="identity",position="dodge")+
> scale_fill_discrete(name="Model",
> labels=c("Multiple", "Simple"))+
> labs(title =paste('La différences des coefficients
> entre la régression multiple et simple'),
> x="Models",y="Coefficient")+
> coord_flip()
>
>
>
>
> __
> R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
> 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.
--
Peter Dalgaard, Professor,
Center for Statistics, Copenhagen Business School
Solbjerg Plads 3, 2000 Frederiksberg, Denmark
Phone: (+45)38153501
Office: A 4.23
Email: pd@cbs.dk Priv: pda...@gmail.com
__
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
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.
Re: [R] Problem of intercept?
Sigh...
In a linear model with qualitative predictor variables, models with and
without intercepts are just different parameterizations of the *same*
model. -- they produce exactly the same predicted responses. So what do
you mean?
Search on "contrasts in linear models R" and similar for an explanation.
Cheers,
Bert
On Tue, Feb 21, 2023, 13:34 varin sacha via R-help
wrote:
> Dear R-experts,
>
> Here below my R code working with quite a few warnings.
> x11 and x12 are dichotomous variable (0=no and 1=yes). I substract 1 to
> ignore intercept.
> I would like not to ignore intercept. How to modify my R code because if I
> just remove -1 it does not work?
>
>
> y= c(32,45,65,34,23,43,65,76,87,98,7,867,56,45,65,76,88,34,55,66)
> x11=c(0,1,1,0,0,1,1,1,0,0,1,0,0,1,0,0,1,1,0,1)
> x12=c(0,1,0,1,0,1,1,0,1,1,0,0,1,1,1,0,0,1,0,0)
>
> Dataset=data.frame(y,x11,x12)
>
> a=lm(y~x11+x12-1,Dataset)$coef
> b=NULL
> for(i in c(1:2)) {
> f=formula(paste('y~',names(Dataset)[i],-1))
> b=c(b,lm(f,Dataset)$coef)
> }
> coef=data.frame(rbind(a,b))
> coef$Model=c('Multi','Single')
> library(reshape2)
> coef.long<-melt(coef,id.vars="Model")
>
> library(ggplot2)
> ggplot(coef.long,aes(x=variable,y=value,fill=Model))+
> geom_bar(stat="identity",position="dodge")+
> scale_fill_discrete(name="Model",
> labels=c("Multiple", "Simple"))+
> labs(title =paste('La différences des coefficients
> entre la régression multiple et simple'),
> x="Models",y="Coefficient")+
> coord_flip()
>
>
>
>
> __
> R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
[[alternative HTML version deleted]]
__
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
[R] Problem of intercept?
Dear R-experts,
Here below my R code working with quite a few warnings.
x11 and x12 are dichotomous variable (0=no and 1=yes). I substract 1 to ignore
intercept.
I would like not to ignore intercept. How to modify my R code because if I just
remove -1 it does not work?
y= c(32,45,65,34,23,43,65,76,87,98,7,867,56,45,65,76,88,34,55,66)
x11=c(0,1,1,0,0,1,1,1,0,0,1,0,0,1,0,0,1,1,0,1)
x12=c(0,1,0,1,0,1,1,0,1,1,0,0,1,1,1,0,0,1,0,0)
Dataset=data.frame(y,x11,x12)
a=lm(y~x11+x12-1,Dataset)$coef
b=NULL
for(i in c(1:2)) {
f=formula(paste('y~',names(Dataset)[i],-1))
b=c(b,lm(f,Dataset)$coef)
}
coef=data.frame(rbind(a,b))
coef$Model=c('Multi','Single')
library(reshape2)
coef.long<-melt(coef,id.vars="Model")
library(ggplot2)
ggplot(coef.long,aes(x=variable,y=value,fill=Model))+
geom_bar(stat="identity",position="dodge")+
scale_fill_discrete(name="Model",
labels=c("Multiple", "Simple"))+
labs(title =paste('La différences des coefficients
entre la régression multiple et simple'),
x="Models",y="Coefficient")+
coord_flip()
__
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
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.
