Posted on by

データ解析基礎論A グラフの基礎

データ解析基礎論A 第2週　グラフの基礎

# descriptive statistics
dat<-read.table("http://www.matsuka.info/data_folder/aov01.txt",header=T)
summary(dat)
mean(dat$shoesize)
var(dat[,1:2])
cov(dat[,1:2])
cor(dat[,1:2])

# basics - plotting
x=seq(-3,3,0.1);y=x^2;
plot(x,y)
plot(x,y,col='red')
plot(x,y,pch=20)
plot(x,y,type='l')
plot(x,y,type='l',lty=4,lwd=3)
plot(x,y,main="THIS IS THE TITLE", xlab="Label for X-axis",ylab="Label for Y-axis")
plot(x,y,main="TITLE", xlab="X here",ylab="Y here",xlim=c(-3.5,3.5),ylim=c(-0.5, 10))
plot(x,y,col='blue',type='o',lty=2,pch=19,lwd=3,main="Y=X*X", xlab="X",ylab="X*X",
  xlim=c(-3.5,3.5),ylim=c(-0.5, 10))

# histogram
dat<-read.table("http://www.matsuka.info/data_folder/aov01.txt",header=T)
hist(dat$h,main="Histogram of Height",xlab="Height",col='blue',xlim=c(140,190))
dens<-density(dat$h);
hist(dat$h,main="Histogram of Height",xlab="Height",xlim=c(140,190),probability=T)
lines(dens,lwd=2,col='red',lty=2)

# boxplot
boxplot(dat$h,main="Boxplot of Height",ylab="Height",col='cyan',ylim=c(140,190))
boxplot(dat$h~dat$gender,main="Distribution of Height by Gender",ylab="Gender",
 xlab="Height",col=c('blue','cyan'),ylim=c(140,190),horizontal=T)
boxplot(dat$h~dat$gender+dat$affil, main="Distribution of Height by Gender and Affiliation",
  ylab="Gender x Affiliation", xlab="Height", col=c('blue','cyan','red','magenta'),
  ylim=c(140,190),horizontal=T)

# barplot
install.packages("gplots")
library(gplots)

means <- tapply(dat$h, dat$gender, mean)
sds<-tapply(dat$h,dat$gender,sd)     
ns<-tapply(dat$h,dat$gender,length) 
sems = sds/sqrt(ns) 
barplot2(means, plot.ci=T,
 ci.l = means - sems,
 ci.u = means + sems,
 ylim = c(140,180),
 names.arg = c("Female","Male"),
 xpd = F,
 ylab = "height",
 xlab = "gender")

# another barplot 
means <- tapply(dat$h,list(dat$gender,dat$affil),mean)
sds <- tapply(dat$h,list(dat$gender,dat$affil),sd)
ns <- tapply(dat$h,list(dat$gender,dat$affil),length)
sem = sds/sqrt(ns)
barplot2(means[1:4], plot.ci=T, ci.l=means[1:4]-sem[1:4],   
        ci.u=means[1:4] + sem[1:4], ylim=c(150,175), 
    names.arg=c("Female,CS","Male,CS","Female,PSY","Male,PSY"),   
        xpd=F,ylab="height",xlab="gender & affiliation")

# histogram again
par(mfrow=c(1,2))
hist(dat[dat$gender=='F',]$h,main="Dist. of Height for Female Participants",
 xlab="Height",xlim=c(140,190),probability=T)
dens.F=density(dat[dat$gender=='F',]$h);lines(dens.F,col='blue',lwd=2)
hist(dat[dat$gender=='M',]$h,main="Dist. of Height for Male Participants",
 xlab="Height",xlim=c(140,190),probability=T,ylim=c(0,0.08))
dens.M=density(dat[dat$gender=='M',]$h);lines(dens.M,col='green',lwd=2)

par(mfrow=c(2,1))

par(mfrow=c(1,1))
plot(dens.F,col='blue',lwd=2,main="Dist. of Height by gender",xlab='Height',
  ylab='density',xlim=c(140,190))
lines(dens.M,col='green',lwd=2)
legend("topleft", c('Female','Male'),col=c('blue','green'),cex=1.5,lwd=2)

# inserting text
text(157.5,0.04,'Female',col='blue',cex=2)
text(170,0.04,'Male',col='green',cex=2)

# scatterplot
plot(dat$shoesize,dat$h,main="Relationship b/w shoesize and height",xlab='shoe size', 
  ylab='height',pch=19,col='red')
text(22,175,paste("r =",substr(cor(dat$shoesize,dat$h),1,5)),cex=1.5)
abline(h=mean(dat$h),col='blue');
abline(v=mean(dat$shoesize),col='green');
text(21.5,165,'mean height',col='blue')
text(25.7,145,'mean shoesize',col='green')
abline(lm(dat$h~dat$shoesize),lty=2,lwd=2)

plot(dat[dat$gender=='F',]$shoesize,dat[dat$gender=='F',]$h,
 main="Relationship b/w shoesize and height",xlab='shoesize',ylab='height',
 cex.lab=1.5,pch=19,col='blue',xlim=c(20,29),ylim=c(140,190))
lines(dat[dat$gender=='M',]$shoesize,dat[dat$gender=='M',]$h,type='p',pch=15,col='green')
legend("topleft", c('Female','Male'), pch =c(19,15),col=c('blue','green'),cex=1.5)
dat.reg<-read.csv("http://www.matsuka.info/data_folder/tdkReg01.csv", header=T)
plot(dat.reg,pch=20,col=c('blue'))
dat.pca<-read.table("http://www.matsuka.info/data_folder/tdkPCA01.txt",header=T)

# intro to central limit theorem
ckCLT=function(n_rep,n_sample){
  dat<-matrix(rnorm(n_rep*n_sample),nrow=n_rep,ncol=n_sample);
  means<-rowMeans(dat)}
n_rep=10^6
n5=ckCLT(n_rep,5)
hist(n5,main="Dist. of sample meanx",xlab="sample mean",xlim=c(-3,3),probability=T)
den5=density(n5);lines(den5,col='blue',lwd=2)

n10=ckCLT(n_rep,10)
n25=ckCLT(n_rep,25)
n100=ckCLT(n_rep,100)
plot(den5,col='blue',lwd=2,,main="Dist. of sample meanx",xlab="sample mean",
  xlim=c(-2,2),ylim=c(0,4))
den10=density(n10);lines(den10,col='red',lwd=2)
den25=density(n25);lines(den25,col='black',lwd=2)
den100=density(n100);lines(den100,col='green',lwd=2)
legend("topleft", c('N=5','N=10','N=25','N=100'),col=c('blue','red','black','green'),
  cex=1.5,lwd=2)

Leave a Reply