df=data.frame(x=1:10,y=c(1,3,2,5,6,8,5,11,13,14))
df
plot(y~x,data=df)
plot(x,y)
plot(df$x,df$y)
lm(y~x,data=df)
plot(y~x,data=df)
df=data.frame(x=1:10,y=c(1,3,2,5,6,8,5,11,13,14))
abline(lm(y~x,data=df))
fit=lm(y~x,data=df)
summary(fit)
coef(fit)
fit=lm(y~x,data=df)
fit$coef
summary(fit)$r.sq
str(summary(fit))
fit$coef
summary(fit)$coef
df=data.frame(x=1:10,y=c(1,3,2,5,6,8,5,11,13,14))
str(df)
str(summary(fit))
df=data.frame(x=1:10,y=c(1,3,2,5,6,8,5,11,13,14))
df$f=c('a','b','a','b','b','a')
df$f=c('a','b','a','b','b','a','b','a','a','b')
df
lm(y~f,data=df)
df$f=as.factor(c('a','b','a','b','b','a','b','a','a','b'))
str(df)
tapply(df$y,df$f,mean)
mean(df$y[df$f=='a'])
mean(df$y[df$f=='b'])
df
lm(y~f,data=df)
tapply(df$y,df$f,mean)
lm(y~x+f,data=df)
ptA=df$f=='a'
ptA
points(y~x,data=df[ptA,],col='red')
points(y~x,data=df[ptA,],col='red',pch=16)
points(y~x,data=subset(df,ptA),col='red',pch=16)
points(y~x,data=subset(df,!ptA),col='blue',pch=16)
abline(lm(y~x,data=subset(df,ptA)),col='red')
abline(lm(y~x,data=subset(df,!ptA)),col='blue')
points(y~x,data=subset(df,!ptA),col='blue',pch=3)
fit=lm(y~x+f,data=df)
fit$coef
plot(y~x,data=df)
points(y~x,data=subset(df,ptA),col='red',pch=16)
abline(fit$coef[1:2],col='red')
points(y~x,data=subset(df,!ptA),col='blue',pch=16)
cfA=fit$coef[1:2]
cfA
abline(cfA,col='blue')
cfB=cfA
cfB[1]+fit$coef[3]
cfB[1]=fit$coef[1]+fit$coef[3]
cfB
fit$coef
abline(cfB,col='red')
fit=lm(y~x+f,data=df)
attach('~/Desktop/treemass.rdata')
ls(2)
search()
ls(2)
ls(2)
head(treemass)
str(treemass)
colnames(treemass)
options(width=80)
colnames(treemass)
options(width=50)
colnames(treemass)
options(width=60)
colnames(treemass)
table(treemass$ForestType)
plot(y~x,data=df)
abline(cfB,col='red')
cf=c(6,0)
abline(cf,col='blue')
df
yhat=cf[1]+df$x*cf[2]
yhat
cfB
yhat=cfB[1]+df$x*cfB[2]
points(df$x,yhat)
points(df$x,yhat,pch=3)
yhat=cf[1]+df$x*cf[2]
points(df$x,yhat,pch=3)
yhat-df$y
(yhat-df$y)^2
sum((yhat-df$y)^2)
yhatB=cfB[1]+df$x*cfB[2]
sum((yhatB-df$y)^2)
sumsq=function(param,x,y) { yhat=x*param[2]+param[1]; return(sum(yhat-y)^2)) }
sumsq=function(param,x,y) { yhat=x*param[2]+param[1]; return(sum(yhat-y)^2) }
sumsq
sumsq(param=cfB,x=df$x,y=df$y)
sumsq(param=c(0,0),x=df$x,y=df$y)
yhatB
sumsq=function(param,x,y) { yhat=x*param[2]+param[1]; return(sum((yhat-y)^2)) }
sumsq(param=cfB,x=df$x,y=df$y)
sumsq=function(param,x,y) { yhat=x*param[2]+param[1];
return(sum((yhat-y)^2)) }
sumsq(param=cfA,x=df$x,y=df$y)
coef(lm(y~x,data=df))
cf=coef(lm(y~x,data=df))
sumsq(param=cf,x=df$x,y=df$y)
sumsq(param=c(-1,1.4),x=df$x,y=df$y)
sumsq(param=c(-1,1.42),x=df$x,y=df$y)
?optim
?optim
?optim
optim(par=c(0,0),fn=sumsq,x=df$x,y=df$y)
optim(par=c(0,0),fn=sumsq,x=df$x,y=df$y,control=list(trace=1))
fit=optim(par=c(0,0),fn=sumsq,x=df$x,y=df$y,control=list(trace=1))
fit=optim(par=c(0,0),fn=sumsq,x=df$x,y=df$y)
fit$val
fit$par
coef(lm(y~x,data=df))
fit=optim(par=c(0,0),fn=sumsq,x=df$x,y=df$y)
?optim
fit=optim(par=c(0,0),fn=sumsq,x=df$x,y=df$y,control=list(reltol=1e-9))
fit$par
fit=optim(par=c(0,0),fn=sumsq,x=df$x,y=df$y,control=list(reltol=1e-10))
fit$par
summary(lm(y~x,data=df))$resid
summary(lm(y~x,data=df))$resid^2
sum(summary(lm(y~x,data=df))$resid^2)
fit=optim(par=c(0,0),fn=sumsq,x=df$x,y=df$y)
fit$val
sumsq=function(param,x,y) { yhat=x*param[2]+param[1];
return(sum((yhat-y)^2)) }
source('~/Desktop/teaching.functions.r')
args(sumsq.linearmodel)
sumsq.linearmodel=function(beta,x,y,graphit=FALSE)
{
 yhat=beta[1]+x*beta[2]
 dev.squared=(yhat-y)^2
 sumsq=sum(dev.squared,na.rm=TRUE)
    
 cat(beta,sumsq,"\n")
 if(graphit) graph.linearmodel(beta,x,y,addpts=FALSE)
# browser()
 return(sumsq)
}
source('~/Desktop/teaching.functions.r')
sumsq.linearmodel(beta=c(0,0),x=df$x,y=df$y)
sumsq(param=c(0,0),x=df$x,y=df$y)
fit=optim(par=c(0,0),fn=sumsq,x=df$x,y=df$y)
fit$par
fit=optim(par=c(0,0),fn=sumsq.linearmodel,x=df$x,y=df$y)
fit=optim(par=c(0,0),fn=sumsq,x=df$x,y=df$y)
save.history('~/meetings_workshops/Rmodeling/Cenpat/historyTuesAM9Oct.r')
savehistory('~/meetings_workshops/Rmodeling/Cenpat/historyTuesAM9Oct.r')
plot(y~x,data=df)
abline(cf,col='blue')
dnorm(3,mean=0,sd=1)
textx=seq(-10,10,by=1)
testx=seq(-10,10,by=1)
testx
dnorm(testx,mean=0,sd=1)
plot(testx,dnorm(testx,mean=0,sd=1))
cumprod(dnorm(testx,mean=0,sd=1))
plot(y~x,data=df)
dnorm(3,mean=0,sd=1)
abline(cf,col='blue')
dnorm(.3,mean=0,sd=1)
dnorm(3,mean=0,sd=1,log=TRUE)
dnorm(.3,mean=0,sd=1,log=TRUE)
plot(testx,dnorm(testx,mean=0,sd=1,log=TRUE))
sumsq
llike=function(param,x,y) { yhat=x*param[2]+param[1];
return() }
plot(y~x,data=df)
llike=function(param,x,y) { yhat=x*param[2]+param[1];
return(sum(dnorm(yhat-y,mean=0,sd=2,log=TRUE))) }
llike
llike(param=c(0,0),x=df$x,y=df$y)
llike(param=c(-1,1.4),x=df$x,y=df$y)
cf
llike(param=cf,x=df$x,y=df$y)
sumsq(param=c(-1,1.4),x=df$x,y=df$y)
sumsq(param=c(0,0),x=df$x,y=df$y)
fit=optim(par=c(0,0),fn=sumsq,x=df$x,y=df$y)
?optim
fit=optim(par=c(0,0),fn=sumsq,x=df$x,y=df$y,control=list(fnscale=(-1))
)
q,x=df$x,y=df$y,control=list(fnscale=(-1))
fit=optim(par=c(0,0),fn=llike,x=df$x,y=df$y,control=list(fnscale=(-1)))
fit$par
llike=function(param,x,y) { yhat=x*param[2]+param[1];
return(sum(dnorm(yhat-y,mean=0,sd=2,log=TRUE))) }
llike=function(param,x,y) { yhat=x*param[2]+param[1];
return(sum(dnorm(yhat-y,mean=0,sd=param[3],log=TRUE))) }
fit=optim(par=c(0,0),fn=llike,x=df$x,y=df$y,control=list(fnscale=(-1)))
fit=optim(par=c(0,0,10),fn=llike,x=df$x,y=df$y,control=list(fnscale=(-1)))
fit$par
abline(fit$par[1:2])
savehistory('~/meetings_workshops/Rmodeling/Cenpat/historyTuesPM9Oct.r')
savehistory('~/meetings_workshops/Rmodeling/Cenpat/historyTuesPM9Oct.txt')