データ解析の前準備 in R

平均やばらつきをみたり，何か解析手法を適用する前にデータ自体をある程度把握しておきたい．
難しそうで，かっこいい手法やるよりも，まずはしっかりと基本的な特徴を眺めてみることも重要．
そこで，一般的にやることをRでさくっとやるためのテンプレート．

# single variant
x = rnorm(n=100, mean=0, sd=3) # test data
summary(x)            # base statistics (mean, median, max, min)
mean(x)               # If you want to get a specific column data.
length(x)             # data size
var(x)                # un-biased variance
sd(x)                 # un-biased standard deviation
mean((mean(x) - x)^2) # sample variance
plot(x, col="blue")   # scatterplot
hist(x, col="blue")   # histogram

# two variant
err = rnorm(n=100, mean=0, sd=1.0)
y = 4.0*x + err
plot(x,y) # scatterplot
cor(y,x)  # corelation
lm(y~x)   # linear regression model

# multi variant
x1 = rnorm(n=100, mean=0, sd=3) # test data
x2 = 3.0*x1 + rnorm(n=100, mean=0, sd=2)
x3 = 3.0*x1 + 2.0*x2 + rnorm(n=100, mean=0, sd=2)
z = data.frame(x1, x2, x3)
cov(z)   # co-variance matrix
cor(z)   # co-relation matrix
paris(z) # scatterplot matrix

# multi variant (categorical) boxplot
sex = c(rep("M", 10), rep("F", 10)) # 10 mans and females
math.score = c(rnorm(10,mean=80,sd=10), rnorm(10,mean=60,sd=5))
eng.score = c(rnorm(10,mean=60,sd=20), rnorm(10,mean=80,sd=5))
ave.score = (math.score+eng.score)/2.0
d = data.frame(sex, math.score, eng.score, ave.score)
boxplot(d$math.score, names=c("MATH"), col="blue") # math socre box
boxplot(d$math.score, d$eng.score, d$ave.score,    # total score box
        names=c("MATH", "ENGLISH", "AVE."), col="blue")
boxplot(d$math.score ~ sex, data=d, col=c("blue", "red"))