Send output to:
Browser Blue - Charts White
Browser Black/White
CSV
Data X:
5.6 4.4 2.4 4.8 3.2 4 4 4.4 6.4 4.4 5.2 4.8 3.2 4.8 4.4 1.6 3.6 3.2 3.2 5.6 6 6.4 3.6 5.6 4.4 3.2 3.6 3.6 3.6 3.6 4 6.4 4.4 3.2 3.6 6.4 4.4 6.4 4.8 4.8 5.6 3.6 4 3.6 4 4.8 5.6 5.6 4 5.6 6.4 3.6 4 2.4 3.2 5.2 4 3.2 2.8 6 3.6 4 4.8 5.2 4 4.4 3.2 3.6 5.2 4.4 3.2 3.6 3.6 6 3.6 4 5.6 4.8 4.8 4.4 5.6 2.4 4.8 3.2 5.6 4.4 4 5.6 4.8 4 5.6 2 4.4 4 3.6 4 6.4 5.2 3.6 4 4 2.8 3.6 3.2 5.6 5.6 3.2 3.6 5.6 5.6 3.2 3.2 3.2 2.8 2.4 3.2 2.4 4.4 5.6 4.4 4.4 4.4 5.6 3.2 8 4.4 3.2 4.4 4 5.6 4.4 3.6 3.6 3.2 4 5.2 5.2 4.8 3.2 5.2 5.6 4.8 5.6 6 5.2 6.4 3.6 3.6 3.6 3.2 2.8 6.4 4.4 3.6 4.4 3.6 5.6 5.2 6.4
Data Y:
5.5 3.5 8.5 5 6 6 5.5 5.5 6 6.5 7 8 5.5 5 5.5 7.5 4.5 5.5 8.5 8.5 5.5 9 7 5 5.5 7.5 7.5 6.5 8 6.5 4.5 9 9 6 8.5 4.5 4.5 6 9 6 9 7 7.5 8 5 5.5 7 4.5 6 8.5 2.5 6 6 3 12 6 6 7 3.5 6.5 6 6.5 7 4 5.5 4.5 5.5 6.5 5 5.5 6 4.5 7.5 9 7.5 6 6.5 7 5 6.5 6.5 5.5 6.5 8 4 8 5.5 4.5 8 6 7 4 4.5 7.5 5.5 10.5 7 9 6 6.5 7.5 6 9.5 7.5 5.5 5.5 5 6.5 7.5 6 6 8 4.5 9 4 6.5 8.5 4.5 7.5 4 3.5 6 7 3 4 8.5 5 5.5 7 5.5 6.5 6 5.5 4.5 6 10 6 6.5 6 6 4.5 7.5 12 3.5 8.5 5.5 8.5 5.5 6 7 5.5 8 10.5 7 10 6.5 5.5 7.5 9.5
Sample Range:
(leave blank to include all observations)
From:
To:
bandwidth of density plot
(?)
Chart options
Label y-axis:
Label x-axis:
R Code
par1 <- as.numeric(par1) library(lattice) z <- as.data.frame(cbind(x,y)) m <- lm(y~x) summary(m) bitmap(file='test1.png') plot(z,main='Scatterplot, lowess, and regression line') lines(lowess(z),col='red') abline(m) grid() dev.off() bitmap(file='test2.png') m2 <- lm(m$fitted.values ~ x) summary(m2) z2 <- as.data.frame(cbind(x,m$fitted.values)) names(z2) <- list('x','Fitted') plot(z2,main='Scatterplot, lowess, and regression line') lines(lowess(z2),col='red') abline(m2) grid() dev.off() bitmap(file='test3.png') m3 <- lm(m$residuals ~ x) summary(m3) z3 <- as.data.frame(cbind(x,m$residuals)) names(z3) <- list('x','Residuals') plot(z3,main='Scatterplot, lowess, and regression line') lines(lowess(z3),col='red') abline(m3) grid() dev.off() bitmap(file='test4.png') m4 <- lm(m$fitted.values ~ m$residuals) summary(m4) z4 <- as.data.frame(cbind(m$residuals,m$fitted.values)) names(z4) <- list('Residuals','Fitted') plot(z4,main='Scatterplot, lowess, and regression line') lines(lowess(z4),col='red') abline(m4) grid() dev.off() bitmap(file='test5.png') myr <- as.ts(m$residuals) z5 <- as.data.frame(cbind(lag(myr,1),myr)) names(z5) <- list('Lagged Residuals','Residuals') plot(z5,main='Lag plot') m5 <- lm(z5) summary(m5) abline(m5) grid() dev.off() bitmap(file='test6.png') hist(m$residuals,main='Residual Histogram',xlab='Residuals') dev.off() bitmap(file='test7.png') if (par1 > 0) { densityplot(~m$residuals,col='black',main=paste('Density Plot bw = ',par1),bw=par1) } else { densityplot(~m$residuals,col='black',main='Density Plot') } dev.off() bitmap(file='test8.png') acf(m$residuals,main='Residual Autocorrelation Function') dev.off() bitmap(file='test9.png') qqnorm(x) qqline(x) grid() dev.off() load(file='createtable') a<-table.start() a<-table.row.start(a) a<-table.element(a,'Simple Linear Regression',5,TRUE) a<-table.row.end(a) a<-table.row.start(a) a<-table.element(a,'Statistics',1,TRUE) a<-table.element(a,'Estimate',1,TRUE) a<-table.element(a,'S.D.',1,TRUE) a<-table.element(a,'T-STAT (H0: coeff=0)',1,TRUE) a<-table.element(a,'P-value (two-sided)',1,TRUE) a<-table.row.end(a) a<-table.row.start(a) a<-table.element(a,'constant term',header=TRUE) a<-table.element(a,m$coefficients[[1]]) sd <- sqrt(vcov(m)[1,1]) a<-table.element(a,sd) tstat <- m$coefficients[[1]]/sd a<-table.element(a,tstat) pval <- 2*(1-pt(abs(tstat),length(x)-2)) a<-table.element(a,pval) a<-table.row.end(a) a<-table.row.start(a) a<-table.element(a,'slope',header=TRUE) a<-table.element(a,m$coefficients[[2]]) sd <- sqrt(vcov(m)[2,2]) a<-table.element(a,sd) tstat <- m$coefficients[[2]]/sd a<-table.element(a,tstat) pval <- 2*(1-pt(abs(tstat),length(x)-2)) a<-table.element(a,pval) a<-table.row.end(a) a<-table.end(a) table.save(a,file='mytable.tab')
Compute
Summary of computational transaction
Raw Input
view raw input (R code)
Raw Output
view raw output of R engine
Computing time
0 seconds
R Server
Big Analytics Cloud Computing Center
Click here to blog (archive) this computation